Regulatory frameworks like the EU GDPR and the EU AI Act approach corporate algorithmic scoring and automated profiling through two distinct, highly complementary lenses.
While the GDPR regulates the data powering the decision and grants individuals structural rights to contest it, the EU AI Act regulates the system itself, imposing hard architectural limits and classification barriers based on how much risk the scoring model poses to human rights.
Here is how both frameworks attempt to check corporate profiling and scoring algorithms:
1. The EU AI Act: Structural Bans and High-Risk Gates
The EU AI Act targets algorithmic scoring by dividing AI systems into strict risk tiers, completely outlawing certain types of profiling while placing heavy compliance burdens on others.
The Absolute Prohibition on Social Scoring (Article 5): The AI Act explicitly bans AI-enabled social scoring systems—whether deployed by governments or private corporations. If a company builds an algorithm that evaluates or classifies individuals or groups over time based on their social behavior, personality traits, or predicted lifestyle, resulting in detrimental or unfavorable treatment in unrelated contexts (e.g., a corporate algorithm tracking consumer habits to arbitrarily deny housing, insurance access, or employment opportunities), it is outlawed outright.
The High-Risk Classification for Credit and HR: Any corporate scoring algorithm used to evaluate creditworthiness, determine credit scores, manage recruitment, filter job applicants, or monitor worker performance is automatically classified as High-Risk.
Mandatory Constraints on High-Risk Scoring: To deploy a credit-scoring or worker-evaluation model legally, companies must implement:
Data Governance: Rigorous training data protocols to ensure the scoring model doesn’t ingest biased variables that produce discriminatory outcomes.
Traceability and Logging: Automatic logging of events and model outputs so auditors can trace why a specific score or denial was generated.
Human Oversight: Designing the system so that a human operator can actively review, override, or reverse automated scores before they cause tangible harm to a citizen.
2. The GDPR: The Right to Explanation and Contestability
While the AI Act governs the design and deployment of the system, the GDPR governs the personal data processed by that system. It targets automated profiling via foundational rights enshrined in Article 22 (and mirrored in national data protection laws).
The Restriction on Solely Automated Decisions: The GDPR traditionally restricts decisions based solely on automated processing (including profiling) that produce legal or similarly significant effects (such as an automated corporate system instantly rejecting a loan or a mortgage).
The Right to Meaningful Information: If a corporation uses automated scoring to judge a consumer, the GDPR forces transparency. Companies cannot hide behind “black-box proprietary algorithms.” They must provide meaningful information about the logic involved, the significance of the profiling, and the anticipated consequences.
Human Intervention on Demand: Under data protection guardrails, individuals targeted by automated corporate profiling have the enforceable right to demand that a human being look at their case, express their point of view, and contest the algorithmic score.
The Convergence: Catching the Corporate Black Box
When combined, these two frameworks dismantle the corporate excuse that “the algorithm decided it.”
If a company deploys an opaque algorithmic scoring model to evaluate human beings, they face a pincer movement: GDPR forces them to open the black box to the individual user and justify the score, while the AI Act forces them to prove beforehand that the scoring architecture is non-discriminatory, heavily logged, supervised by humans, and clear of prohibited “social scoring” traps.
Penalties for non-compliance
Both the EU AI Act and the GDPR use a “whichever is higher” formula for financial penalties—calculating the maximum fine using a flat baseline cash cap or a percentage of a corporation’s total worldwide annual turnover, whichever yields the greater amount.
For multinational technology corporations, the percentage-based global turnover caps dwarf the flat sums, exposing them to billions in potential losses.
1. The EU AI Act Penalties
The EU AI Act structures its maximum penalties into a tiered pyramid based on the severity of the violation. Breaches involving banned practices or data governance failures carry the harshest teeth ever written into European tech regulation.
Tier 1: Prohibited AI Practices (e.g., Social Scoring & Behavioral Manipulation)
Applies to: Deploying banned systems like citizen social-scoring algorithms, subliminal manipulation, or exploitative biometric categorization.
Maximum Penalty:Up to €35 million or 7% of the company’s total worldwide annual turnover (whichever is higher).
Context: This 7% ceiling explicitly exceeds the GDPR’s maximum cap, specifically designed to ensure tech giants cannot absorb fines as a mere cost of doing business.
Tier 2: High-Risk AI Obligations & Transparency Failures
Applies to: Violating compliance mandates for high-risk systems (such as corporate credit-scoring algorithms, recruitment filters, or biometric tools) or failing transparency requirements.
Maximum Penalty:Up to €15 million or 3% of global annual turnover (whichever is higher).
Applies to: Supplying incorrect, incomplete, or misleading information to national competent authorities or notified bodies during audits.
Maximum Penalty:Up to €7.5 million or 1% of global annual turnover (whichever is higher).
2. The GDPR Profiling & Data Protection Penalties
Under GDPR (Article 83), fines are similarly split into two tiers depending on which core rights or principles were violated. Unlawful automated profiling, lack of a lawful basis for tracking consumer behavior, or ignoring data subject rights fall squarely into the higher tier.
Applies to: Breaching fundamental processing principles, running automated profiling without a valid legal basis, violating user rights (Articles 12–22), or executing illegal international data transfers.
Maximum Penalty:Up to €20 million or 4% of total worldwide annual turnover (whichever is higher).
Real-World Precedent: This tier has driven massive historical enforcement actions against major tech firms, such as Ireland’s DPC issuing a €310 million fine to LinkedIn for unlawful behavioral profiling and targeted advertising.
Lower Tier: Administrative & Governance Breaches
Applies to: Technical security failures, failure to maintain adequate records of processing activities, or failing to notify a supervisory authority of a data breach within 72 hours.
Maximum Penalty:Up to €10 million or 2% of total worldwide annual turnover (whichever is higher).
The Cumulative Regulatory Pressure
When corporations deploy automated scoring and profiling systems that cross legal boundaries, they do not just face single-file fines. Regulators routinely issue concurrent penalties—hitting a company simultaneously under GDPR for unlawful personal data profiling and under the AI Act for non-compliant, high-risk algorithmic architecture. For a global enterprise, a systemic failure in its scoring models can trigger combined turnover-pegged penalties reaching up to 11% of global annual revenue.
Auditing the Blackbox
Auditing complex neural networks and proprietary scoring algorithms—the ultimate “black boxes”—presents a massive technical hurdle. If a model consists of hundreds of billions of opaque weights distributed across a high-dimensional vector space, a human regulator cannot simply read the source code to find a violation.
To bypass this, European regulators (operating through the European AI Office, national market surveillance authorities, and independent Notified Bodies) do not just try to reverse-engineer the math line-by-line. Instead, they approach the audit through a combination of structural compliance mandates, systematic behavioural probing, and cryptographic logging.
The auditing toolchain relies on three core operational strategies:
1. White-Box Documentation & Data Governance (Before Deployment)
Under the EU AI Act and GDPR, the easiest way to prove a violation is to inspect what the corporation built before the model ever touched live data. Regulators bypass the black box by forcing companies to open their engineering notebooks.
Data Provenance and Training Audits: Under Article 10 of the AI Act, providers of high-risk scoring systems must prove the datasets used to train the model were representative, clean, and free of systemic bias. Auditors inspect the version-controlled records of data-cleaning steps, gap analyses, and labeling protocols. If a credit-scoring algorithm discriminates against a protected class, regulators trace it straight back to biased training data.
The Technical Documentation Dossier (Annex IV): Companies must legally submit comprehensive documentation detailing the system’s architecture, optimization objectives, hardware requirements, and the logic behind its classification thresholds. If the deployed model behaves differently than what was documented, the provider faces immediate fines for deceptive compliance.
2. Behavioral Probing and Counterfactual Testing (Black-Box Audits)
When regulators or independent researchers test an active system (such as a corporate hiring filter or insurance scoring engine), they rely on behavioral and counterfactual probing.
Instead of looking inside the model, they systematically manipulate inputs and observe the statistical outputs to map the hidden decision boundaries:
The Sock-Puppet Audit: Auditors generate hundreds of synthetic profiles with identical financial or professional credentials—varying only a protected characteristic (such as gender, ethnic indicator, or zip code). If the automated scoring model systematically assigns lower trust scores or loan approvals to one demographic, the statistical disparity serves as empirical proof of algorithmic discrimination.
Adversarial Stress-Testing: Regulators inject edge-case inputs designed to force the model to fail or reveal hidden biases, testing whether the system maintains robustness against manipulation or generates unauthorized profiling metrics.
3. Automated Event Logging and Traceability (The Flight Data Recorder)
Neural networks may be non-deterministic and opaque, but the infrastructure running them is completely deterministic.
Article 12 of the EU AI Act mandates that high-risk AI systems must feature automatic logging capabilities throughout their entire operational lifecycle.
System Telemetry: The hosting servers must maintain secure, immutable logs recording every significant event: inputs received, confidence scores generated, system errors, and triggers for human oversight intervention.
Reconstructing the Decision Path: If a citizen files a grievance claiming an illegal automated profiling decision, regulators do not need to understand every neural weight. They pull the system logs for that specific transaction to reconstruct the exact data state, feature weights, and threshold triggers that produced the negative score.
The Ultimate Regulatory Checkmate
The genius of these frameworks is that they convert an impossible computer science problem (explaining a trillion-parameter neural network) into a strict legal liability framework.
A corporation can no longer hide behind the defense that “the AI is a black box and we don’t know why it made that decision.” Under EU regulation, if a proprietary scoring algorithm produces an unlawful discriminatory outcome or violates profiling restrictions, the opacity of the model is not an excuse—it is a regulatory violation in itself.
Open Source
The European Union’s AI Act approaches open-source foundational models (such as Meta’s Llama or Mistral AI’s open-weight releases) with a specific, highly nuanced distinction: it grants partial exemptions to the model provider based on licensing, but it offers zero immunity to the deployer who puts that model to work.
The regulatory treatment of these models maps directly onto your classification framework. The EU realized that penalizing open-weight models with heavy bureaucratic friction would kill the open-source ecosystem, yet it could not allow powerful models to escape accountability entirely.
1. The Four-Pronged Open-Source Exemption Test
Under Article 53(2) of the AI Act, providers of General-Purpose AI (GPAI) models are granted a partial exemption from certain administrative burdens, but only if their models satisfy four strict conditions:
True Open-Source License: Released under a free and open-source license (e.g., Apache 2.0, MIT) that permits access, usage, modification, and redistribution. (Licenses with “research-only” or restrictive commercial clauses do not qualify).
Publicly Available Weights: The actual model parameters (weights) must be public—”open architecture, closed weights” models do not qualify.
Architectural Transparency: Information concerning the model’s architecture must be publicly released.
Usage Documentation: Clear documentation on model usage must be accessible.
If a model like Mistral or an open-weight Llama variant meets these criteria, the provider is exempt from two major administrative duties:
They do not have to generate and maintain exhaustive internal technical documentation specifically formatted for EU regulatory bodies (Annex XI).
They do not have to supply proprietary downstream documentation packages to every enterprise integrator who builds on top of their model (Annex XII).
2. The Non-Negotiable Baseline (What Open-Source Cannot Escape)
Even if a model is fully open-source and satisfies all four conditions, two obligations can never be waived:
Copyright Compliance: The provider must maintain an explicit policy ensuring compliance with EU copyright law, specifically respecting machine-readable rights reservations (like web-crawler blocks and robots.txt protocols) used during training.
Training Data Summaries: The provider must publish a sufficiently detailed, standardized summary of the content used to train the model.
3. The “Systemic Risk” Ceiling (The Compute Threshold)
The open-source exemption operates as a sliding scale that instantly snaps shut if a model crosses a critical capability threshold.
The AI Act establishes that any GPAI model trained using a cumulative compute power greater than $10^{25}$ FLOPs (Floating-Point Operations) is automatically classified as a Model with Systemic Risk.
When frontier open-source models scale up to or past this threshold, the open-source exemption vanishes entirely.
They are subjected to the full suite of systemic risk obligations: mandatory adversarial red-teaming, rigorous tracking and reporting of serious incidents, cybersecurity evaluations, and structural energy-consumption reporting.
4. The Deployer Trap: Where the Exemption Stops
The most crucial rule of the EU AI Act regarding open-source models is this: The license on the model governs the developer, but the use case governs the deployer.
If a bank, a hospital, or an enterprise downloads an open-weight Llama model from Hugging Face under a completely free Apache 2.0 license and integrates it into a High-Risk AI System (such as credit scoring, biometric categorization, or recruitment filters), the open-source nature of the underlying model provides zero legal protection.
The enterprise deploying the model inherits the full weight of the High-Risk obligations:
They must establish rigorous risk-management systems.
They must guarantee data governance and mitigate algorithmic bias.
They must ensure immutable event logging.
They must bake in active human oversight.
The Verdict on Open-Source Regulation
The EU AI Act treats open-source foundational models as raw infrastructure—similar to how a traditional legal system treats a public highway or a block of steel. The person who mines the steel (the model creator) gets a break on documentation, but the person who builds a vehicle out of it and drives it on public roads (the deployer) is held strictly accountable for its safety.
Model Tuning
If an enterprise fine-tunes an open-source model like Llama for internal use, does that enterprise legally become the ‘provider’ of a new AI system under the EU AI Act?
Under the EU AI Act, the short answer is no, not usually—standard internal fine-tuning does not automatically make an enterprise the “provider” of a General-Purpose AI (GPAI) model. For most routine customisations, the enterprise remains legally classified as a deployer.
However, the law establishes a precise, mathematical boundary where minor customisation ends and “substantial modification” begins.
The regulatory test governing whether an enterprise fine-tuning an open-source model like Llama inherits provider obligations relies on the following criteria:
1. The Standard Rule: Fine-Tuning is Not “Developing”
The European Commission’s guidelines clarify that adapting, prompting, quantising, or performing standard parameter-efficient fine-tuning (like LoRA or standard instruction-tuning) on an existing open-source model does not make you the model provider.
If your internal fine-tuning falls within the scope of what the upstream creator (e.g., Meta) originally anticipated or permitted in their technical documentation, you are treated as a downstream deployer. Your legal duties are limited to using the model responsibly, ensuring human oversight if deployed in a high-risk context, and respecting transparency rules—you do not have to recreate upstream GPAI technical documentation or training data summaries.
2. The Exception: The “One-Third” Compute Rule (Substantial Modification)
The line between a deployer modifying a model and becoming a new provider is measured by computational weight—specifically, floating-point operations (FLOPs):
If your fine-tuning process consumes cumulative compute resources that exceed one-third (33%) of the original model’s base pre-training compute, the EU AI Act presumes you have substantially altered the model.
For a massive model like Llama, 33% of its original pre-training compute is an astronomical amount of energy. Routine, targeted enterprise fine-tuning on internal clusters rarely comes close to touching this threshold.
If an enterprise does cross that one-third compute threshold through heavy, foundational retraining, it legally crosses the boundary and becomes a GPAI provider for that newly modified version.
3. The “Internal Use” Catch (The Deployment Context)
Even if your fine-tuning stays well below the compute threshold and you avoid becoming a GPAI model provider, the intended use of that fine-tuned model still dictates your legal reality.
If an enterprise fine-tunes Llama entirely for internal use (e.g., an internal document search or code assistant), it avoids many external-facing burdens. However, if that same fine-tuned internal model is integrated into a High-Risk AI System (such as an automated recruitment filter screening incoming job resumes, or an internal credit-scoring tool for applicants), the enterprise instantly inherits all strict High-Risk deployer obligations under the Act—regardless of whether it fine-tuned the model or downloaded it straight off Hugging Face.
When an enterprise takes a fine-tuned, open-source model (like a custom version of Llama) and deploys it in a high-risk context (such as automated recruitment, credit scoring, biometric identification, or essential public services), it triggers Article 26 of the EU AI Act.
Because the enterprise is acting as a deployer rather than the original foundation model provider, its duties shift from building the architecture to governing its operational safety, oversight, and traceability.
The specific documentation and risk-management duties mandated by the EU framework include:
1. Fundamental Rights Impact Assessment (FRIA)
Before putting the fine-tuned model live in a high-risk scenario, certain deployers (including private entities providing public services or operating in sensitive sectors like credit scoring and insurance pricing) must conduct and document a Fundamental Rights Impact Assessment (FRIA).
What it requires: A formal evaluation mapping out how the AI model will impact the fundamental rights (e.g., non-discrimination, privacy, worker dignity) of the individuals it interacts with.
Administrative duty: This assessment must be registered in the EU database before the system is put into service.
2. Mandatory Human Oversight (The “Kill Switch” Mandate)
Deployers cannot let a high-risk fine-tuned model operate autonomously without human-in-the-loop safeguards.
What it requires: The enterprise must assign natural persons to oversee the system who possess the necessary competence, training, authority, and support.
Operational duty: Overseers must be positioned to fully understand the model outputs, disregard or override automated decisions when necessary, and have the technical ability to halt or pause the system instantly if an anomaly or systemic bias appears.
3. Rigorous Input Data Governance
Even if the upstream open-source model was trained on general internet data, the enterprise controls the local input data fed into the fine-tuned model during deployment.
What it requires: To the extent the enterprise exercises control over the input data, it must ensure that data is relevant, representative, and cleansed of historical biases that could trigger discriminatory automated profiling.
4. Automated Event Logging (The Flight Recorder)
Neural networks are non-deterministic, but the infrastructure running them must be fully auditable.
What it requires: Under Article 26(6), deployers must ensure that the logs automatically generated by the high-risk AI system are retained for a minimum of six months (unless superseded by sector-specific financial or data protection laws like GDPR).
Auditing duty: These logs must be kept under the enterprise’s control so that regulators or auditors can reconstruct the exact feature weights, inputs, and thresholds that triggered a specific decision.
5. Post-Market Monitoring and Incident Reporting
Deployers cannot simply launch a fine-tuned model and walk away.
What it requires: The enterprise must continuously monitor the operation of the system based on the provider’s instructions for use.
The Escalation Trigger: If the enterprise identifies a serious incident (e.g., a catastrophic bias cascade, systematic discrimination, or a failure threatening fundamental rights), it must immediately suspend use of the system and notify both the original provider and the relevant national market surveillance authority.
6. Transparency and Worker Notification
If the fine-tuned model is deployed internally for workforce management or recruitment:
What it requires: The enterprise must explicitly inform workers’ representatives and affected employees before they are subjected to the high-risk AI system, adhering strictly to labor consultation rules.
AI & GDPR compliance
The intersection between the EU AI Act’s Fundamental Rights Impact Assessment (FRIA) (Article 27) and the GDPR’s Data Protection Impact Assessment (DPIA) (Article 35) represents one of the most critical compliance overlaps for enterprises deploying AI systems.
Because almost every high-risk AI application (like automated recruitment, credit scoring, or worker monitoring) processes personal data, enterprises routinely find themselves triggering both assessments simultaneously.
Rather than treating them as isolated silos, the European Union designed the frameworks to interact through structural bridges, overlapping scopes, and explicit legal linkages.
1. Scope and Focus: Data Privacy vs. Total Human Rights
To understand how they intersect, you first have to look at what each assessment is built to evaluate:
The GDPR DPIA (Article 35): Narrower, deeper, and strictly focused on information privacy and data protection rights. It asks: How does processing personal data impact an individual’s privacy, data security, and informational self-determination? It evaluates necessity, proportionality, data minimization, and technical safeguards.
The AI Act FRIA (Article 27): Broader and focused on holistic fundamental rights. It asks: How does the deployment of this automated system impact human dignity, non-discrimination, worker rights, freedom of expression, and access to essential services? Data privacy is just one small slice of a FRIA.
2. The Direct Legal Bridge (Article 27(4))
The EU explicitly anticipated the administrative nightmare of forcing companies to run two entirely separate bureaucratic processes for the same software.
Under Article 27(4) of the AI Act, the law provides a legal reuse mechanism:
If an enterprise has already conducted a DPIA under Article 35 of the GDPR, it can re-use and integrate those findings directly into its FRIA.
Because a DPIA already maps out data flows, system logic, and data-privacy risks, it serves as the foundational data-architecture chapter of the broader Fundamental Rights Impact Assessment.
3. Key Differences in Operational Requirements
While they can be merged or cross-referenced, a standard GDPR DPIA cannot simply be copy-pasted to satisfy a FRIA. A FRIA demands metrics that a data protection assessment completely ignores:
Requirement
GDPR DPIA (Article 35)
EU AI Act FRIA (Article 27)
Primary Trigger
High risk to data privacy/freedoms via personal data processing.
Deployment of a High-Risk AI system (regardless of underlying data nuances).
Core Focus
Lawfulness, minimization, security, and storage limits of personal data.
Societal harm, systemic bias, socio-economic exclusion, and discrimination.
Human Oversight
Evaluates if staff are trained on privacy rules.
Maps out operational protocols for the human “kill-switch”—ensuring overseers can actually override or halt the AI.
Stakeholder Mapping
Maps data subjects.
Explicitly maps vulnerable groups or communities likely to be impacted by the system’s decisions.
Time/Frequency Scope
Focuses on data life-cycle and retention periods.
Focuses on the duration, frequency, and operational context of the system’s active deployment.
4. The Unified Compliance Workflow in Practice
In an enterprise environment, compliance officers do not treat these as two separate tasks. Instead, they run an Integrated Impact Assessment structured around the following timeline:
Phase 1 (The GDPR Base): The data protection officer (DPO) maps out the data flows, ingestion pipelines, storage limits, and privacy risks, completing the DPIA.
Phase 2 (The AI Act Extension): The compliance team takes that data-flow foundation and expands it into a FRIA. They evaluate how the model’s outputs affect human dignity, test for demographic bias (counterfactual probing), outline the exact procedures for human oversight, and identify vulnerable groups who might be harmed by a false classification.
Phase 3 (Filing and Registration): For high-risk systems, the completed impact assessment package (incorporating both the DPIA data metrics and the FRIA fundamental rights analysis) is logged into the EU database or made available to national market surveillance authorities upon request.
By merging the two, enterprises avoid redundant paperwork while satisfying the dual demand of protecting both data privacy (GDPR) and systemic human rights (AI Act).
On the Probability, Semiotics, and Politics of Detecting Extraterrestrial Intelligence
I. The Universal Language That Isn’t
There is a story we tell ourselves about first contact, and it goes like this: mathematics is the universal language. Any sufficiently advanced civilisation, regardless of its biology, its sensory apparatus, its evolutionary history, will have discovered the same primes, the same geometric constants, the same physical relationships. If we want to speak to the stars, we encode mathematics, because mathematics is what we share. It is the lingua franca of the cosmos, the one thing that is true everywhere and independent of the speaker.
This is a cultural claim dressed as a scientific one. It is, more precisely, a semiotic claim, a claim about the relationship between signs, meaning, and minds, and it has rarely been examined as one.
The discipline that would examine it is semiotics: the study of how meaning is made, transmitted, and received. Semiotics, in the tradition of Charles Sanders Peirce, holds that a sign requires three things, a sign vehicle (the physical form of the sign, the signal itself), an object (the thing the sign refers to), and an interpretant (the meaning constructed by the receiver). A sign is not a sign in itself; it is a sign only in the act of interpretation. Meaning does not reside in the signal. It is constructed, by the receiver, from the signal, using the only tool available: context.
This matters more than it might seem. When we say that mathematics is universal, we are making a claim about the sign vehicle, that the same mathematical structures will be independently discovered by any intelligence. This is plausible, though not as certain as we like to think; even mathematics, as a human practice, is shaped by culture, by language, by the contingent history of the species that produced it. The ancient Greeks did geometry differently from the medieval Islamic mathematicians, who did it differently from the calculus tradition that emerged in Europe. These are not different mathematics, the relationships are the same, but they are different semiotic systems, different ways of encoding and interpreting the same underlying structures. If mathematics were truly transparent, if the sign vehicle mapped unambiguously to the object, then mathematical notation would not have a history. It does.
But the deeper problem is not at the level of the sign vehicle. It is at the level of the interpretant. Even if an extraterrestrial intelligence has discovered the same mathematical relationships, the meaning they construct from a signal encoding those relationships depends on their context, their biology, their sensory world, their temporal experience, their cognitive architecture. A signal that encodes prime numbers, for us, means “intelligence, because primes are a non-natural pattern that requires a mind to generate.” This interpretation depends on a chain of assumptions: that the receiver recognises primality as a non-natural pattern; that they associate non-natural patterns with intelligence; that they associate intelligence with communication; that they associate communication with the intent to be understood. Each of these is a cultural inference, grounded in the specific semiotic world of a species that evolved on a particular planet with a particular evolutionary history.
The Estonian biologist Jakob von Uexküll coined the term Umwelt to describe the perceptual world inhabited by an organism, the specific slice of reality available to it through its particular sensory and cognitive apparatus. A tick’s Umwelt consists of temperature, light, and the smell of butyric acid. A bat’s Umwelt is built from echolocation. A human’s Umwelt is built from vision, hearing, and a peculiar cognitive architecture that is, as far as we know, unique: the capacity for recursive symbolic thought, which has produced language, mathematics, and the stories we tell about the stars.
The assumption underlying SETI, the Search for Extraterrestrial Intelligence, is that the Umwelt of a technological alien species would overlap sufficiently with ours that a signal meaningful in one would be meaningful in the other. This is an assumption, not a fact. We have no evidence for it, because we have no data. We have one data point: ourselves. And the one data point we have is not a sample; it is a case study, and a case study of one tells you nothing about the distribution of the population from which it is drawn.
II. The Imperative of Life
There is a second assumption, deeper and less examined than the first, that underlies the entire enterprise. It is the assumption that life, once it arises, tends toward intelligence, and that intelligence, once it arises, tends toward communication, and that communication, once it is possible, tends toward the stars. This is a narrative of progress, a kind of cosmic teleology, and it is, in its structure, remarkably similar to the Great Chain of Being, the medieval concept that all of creation is arranged in a hierarchy from the lowest to the highest, with humanity (naturally) near the top. We have secularised the Great Chain. We no longer place God at the top. We place technological civilisation there instead, and we assume that the universe, if it contains life at all, will produce civilisations like ours, because ours is the natural endpoint of the chain.
This is not what the observable facts suggest.
The observable facts are these. Life on Earth arose approximately 3.8 billion years ago, possibly earlier. For the first 2.5 billion years of that history, more than half the planet’s lifetime, life consisted of single-celled organisms. Multicellularity arose perhaps 600 million years ago. Complex multicellular life, with differentiated tissues and organs, arose perhaps 550 million years ago. The first animals appeared in the Cambrian, roughly 540 million years ago. From the Cambrian to the present, life on Earth has experienced at least five major mass extinction events, each of which reset the trajectory of evolution and eliminated a significant fraction of existing species.
Intelligence, in the sense relevant to SETI, which is to say, intelligence capable of building technology capable of transmitting or receiving interstellar signals, has arisen exactly once in 3.8 billion years. Not twice. Not in multiple lineages. Once. In a single species, Homo sapiens, which has existed for approximately 300,000 years and has been technologically capable of radio transmission for approximately 120 years.
Let us sit with those numbers. 3.8 billion years of life. One instance of the kind of intelligence we are searching for. 120 years of the kind of technology we are searching with. The fraction of life’s history on this planet during which it has been detectable by the methods SETI uses is approximately:
120 / 3,800,000,000 ≈ 3.2 × 10⁻⁸
Three ten-millionths of the time life has existed. And that is the fraction for our own planet, where we know life exists. For any other planet, we do not even know whether life exists, let alone intelligence, let alone technology.
This is not an argument that life is rare. It is an argument that the specific thing SETI is searching for, a technological civilisation, producing detectable electromagnetic signals, during the narrow window in which we are listening, is, on the evidence available to us, vanishingly improbable. Not impossible. Not absent. But the probability is not a number we can calculate, because we have one data point, and one data point cannot populate a distribution.
What we can say is this: the assumption that intelligence is convergent, that it arises independently in many lineages the way that eyes have evolved independently dozens of times, is not supported by the evidence. On Earth, we have many examples of social complexity (eusocial insects, cetaceans, elephants, corvids, cephalopods), many examples of tool use (chimpanzees, crows, otters, dolphins), many examples of cognitive sophistication (parrots, octopuses, pigs, dogs). We have exactly one example of technology-building intelligence. Eyes have evolved independently because light is a physical constant and the selective advantage of detecting it is overwhelming. Technology-building intelligence is not eyes. It is a specific, contingent, historically bounded phenomenon that arose in a specific lineage under specific conditions, and the evidence that it is a convergent adaptation, that it arises reliably, like vision, wherever the conditions permit, is, at present, zero.
This does not mean it is rare. It means we do not know. And the not-knowing is the honest position, and the not-knowing is the position that the SETI industry, as a matter of institutional logic, cannot afford to occupy.
III. The Signal Problem, What We Are Actually Looking For
Setting aside the semiotic and philosophical questions, let us consider the physics. What would it actually take to detect an extraterrestrial signal?
The problem has three components: attenuation, noise, and coincidence.
Attenuation
Electromagnetic radiation propagates according to the inverse square law. A signal that leaves a transmitter with power P will arrive at a distance d with power density P / (4πd²). At interstellar distances, this is a brutal fact. A signal transmitted from a star 6,000 light-years away, roughly the distance to the Cygnus arm, a reasonable estimate for the scale of any plausible target, arrives at Earth with a power density that is a factor of roughly 10⁻¹⁵ of its transmitted value. To detect a signal at that distance, the transmitter must either emit with enormous power or beam the signal tightly in our direction.
A tight beam is energy-efficient but covers a tiny fraction of the sky. The beamwidth of a transmitting antenna is inversely proportional to the ratio of the antenna’s diameter to the wavelength. To beam a signal at 1.42 GHz (the hydrogen line, a popular SETI frequency) to a target the size of the Earth’s orbit from a transmitter 6,000 light-years away requires an antenna with an effective diameter of approximately 500,000 kilometres, larger than any structure any civilisation we can imagine could build, unless it is distributed across a solar system. And even then, the beam would only reach the Earth if the transmitter knew exactly where the Earth was and was pointed at it with precision.
The alternative is an omnidirectional beacon, a signal broadcast in all directions, detectable from any point in the sky. But an omnidirectional signal at interstellar distances requires power that beggars the imagination. A signal detectable by the Arecibo telescope (when it existed) at 1,000 light-years, broadcast omnidirectionally, would require a transmitter power of approximately 10¹⁵ watts, roughly 100 times the total current electrical generating capacity of human civilisation. At 6,000 light-years, the requirement scales by a factor of 36, to roughly 3.6 × 10¹⁶ watts, a significant fraction of the total solar output intercepted by the Earth.
This is not impossible. A civilisation that has access to the energy of a star, a Kardashev Type II civilisation, which is the kind that science fiction routinely assumes, could, in principle, broadcast omnidirectionally at interstellar distances. But we are now no longer doing science. We are doing speculation, and the speculation is unconstrained by any observation, which means it is, in Karl Popper’s sense, unfalsifiable, which means it is not, strictly speaking, science at all.
The attenuation problem means that the vast majority of plausible transmission scenarios are undetectable. A civilisation that is broadcasting toward us, with a beam pointed at our solar system, using a power level we could detect, at a frequency we are monitoring, at a time we are listening, is a civilisation that is, by construction, very similar to us in its assumptions about how communication works. We are, in effect, searching for ourselves.
Noise
The universe is loud. The cosmic microwave background provides a floor of approximately 2.7 K of thermal noise at all frequencies. Galactic synchrotron radiation adds frequency-dependent noise that is particularly strong at lower frequencies. Earth’s atmosphere emits thermal noise. The receiver itself contributes thermal noise, quantified by its system temperature. And, increasingly, our own civilisation contributes radio frequency interference, satellites, radar, broadcast transmitters, industrial equipment, that contaminates the very frequencies we are searching for.
The signal-to-noise ratio determines detectability. A signal is detectable only if it is significantly stronger than the noise floor integrated over the observation time. For a narrowband signal, the kind SETI traditionally searches for, because a narrowband signal is an indicator of engineering, of a transmitter that has been deliberately designed, the noise in a single frequency bin is proportional to the system temperature divided by the integration time. Longer observations reduce noise, but longer observations cost more telescope time, and telescope time is the scarcest resource in the SETI enterprise.
The “water hole”, the band of frequencies between the hydrogen line (1.42 GHz) and the hydroxyl line (1.66 GHz), is relatively quiet, because galactic background noise decreases with frequency and the atmosphere is transparent in this range. The choice of the water hole is, however, a semiotic choice: it assumes that an alien civilisation would choose to broadcast in this band for the same reasons we would choose to listen in it. The reasoning is that hydrogen and hydroxyl are the components of water, and water is the solvent of life (as we know it), and therefore the band between them is a “cosmic meeting place”, a frequency band that any water-based life would find significant.
This is a lovely idea. It is also, in its structure, an argument from analogy: we find the hydrogen line significant because hydrogen is the most abundant element in the universe and its spectral line is a natural reference frequency. We assume that an alien intelligence would find it significant for the same reasons. But the significance is not in the frequency; it is in the interpretation. And the interpretation depends on the interpretant, which depends on the Umwelt, which depends on the species.
Temporal Coincidence
A civilisation must be transmitting during the window in which another civilisation is listening. We have been listening, in one form or another, for approximately 60 years. If technological civilisations have lifetimes of, say, 10,000 years (a number chosen for illustrative purposes, because we have no basis for choosing any number), and if they arise at some rate per galaxy per unit time (a rate we cannot estimate because we have one data point), then the probability of temporal overlap depends on the ratio of the combined listening-and-transmitting window to the total time available. If both civilisations are rare and short-lived, the probability of overlap is small. If both are common and long-lived, it is large. We do not know which, and the range of plausible values spans many orders of magnitude.
The one observable fact we have is that we have been listening for 60 years and have heard nothing. This is a null result, and null results in science are informative but notoriously difficult to interpret. A null result can mean the phenomenon does not exist. It can mean the phenomenon exists but is below the detection threshold. It can mean the phenomenon exists and is detectable but we are looking in the wrong place, at the wrong frequency, at the wrong time, or with the wrong assumptions. The history of science is rich with null results that were, in retrospect, failures of imagination rather than failures of the phenomenon, the Michelson-Morley experiment, the early null results in the search for exoplanets, the decades of null results in gravitational-wave detection before LIGO.
But the history of science is also rich with null results that were, in retrospect, correct: the search for the ether, the search for Vulcan, the search for N rays. The difference between a null result that precedes a detection and a null result that confirms an absence is visible only in retrospect, and we are not in retrospect. We are in the present, and the present is ambiguous.
IV. The Semiotic Trap
Here is the trap, and it is a semiotic one: we cannot search for a signal without a model of what a signal looks like. And our model of what a signal looks like is, inescapably, a model of what our signal would look like. We search for narrowband carriers because we build narrowband carriers. We search in the water hole because we find the water hole significant. We search for prime numbers because we see primes as a signature of intelligence. We search for structured, repetitive, obviously non-natural patterns because structured, repetitive, obviously non-natural patterns are what we produce when we want to be noticed.
The logic is circular, and the circularity is not a flaw that can be engineered away. It is structural. To search is to assume. To assume is to project. To project is to search for yourself.
This is not an argument against searching. It is an argument for understanding what searching means, and for being honest about the fact that a null result in a search conditioned on a model that is derived from a sample of one tells you only that the model’s specific predictions are not confirmed. It does not tell you that the phenomenon is absent. It does not tell you that the model is wrong. It tells you nothing, in the strict statistical sense, because the prior is unconstrained and the likelihood is dominated by assumptions that are not testable.
The semiotic literature has a term for the specific kind of error this produces: abduction, as defined by Peirce, is the process of inferring the best explanation for an observation. SETI is not, in practice, abductive; it is hypothetical-deductive. We hypothesise a model of alien communication, deduce what the signal would look like, and search for it. When we do not find it, we do not revise the hypothesis, we refine the search. We build more sensitive receivers. We search more frequencies. We observe more stars. The hypothesis, that a technological civilisation would produce a signal recognisable to us, is never tested, because it cannot be tested. It is the precondition of the search, not a result of it.
The Italian semiotician Umberto Eco, in his work on the limits of interpretation, argued that a text can be interpreted in an unbounded number of ways, but not all interpretations are equally valid. The constraint on interpretation is the text itself, the sign vehicle, but also the community of interpreters, the shared context that makes some readings plausible and others not. For SETI, the “text” is the signal (if one exists), and the “community of interpreters” is the scientific community of Earth. The constraint on our interpretation is our own semiotic world, our physics, our mathematics, our engineering, our cultural assumptions about what communication is and why beings communicate. We cannot step outside this constraint. We can only be aware of it.
V. The One Data Point
Let us return to the observable facts, because they are all we have, and they deserve more respect than they typically receive.
We know that life exists on at least one planet. We know that, on that planet, life has produced one technological civilisation. We know that that civilisation has been capable of radio transmission for approximately 120 years. We know that, during those 120 years, the civilisation has not deliberately broadcast a continuous, high-power, omnidirectional signal intended for reception by other stars. (We have sent short, targeted messages, the Arecibo message in 1974, the more recent Breakthrough Message competition, but these are not continuous beacons and would not be detectable by a civilisation like ours at interstellar distances.)
This last fact is, in the context of SETI, almost never discussed. If we are the model for what a technological civilisation does, then the model predicts that a technological civilisation does not broadcast. It listens. It searches. It does not transmit, or it transmits only sporadically and with low power, in short bursts, toward specific targets.
The implication is uncomfortable. If the one civilisation we know does not behave the way SETI assumes civilisations behave, i.e., continuously broadcasting detectable signals, then the search is predicated on a model that is contradicted by the only available data. We are searching for a behaviour that we ourselves do not exhibit.
There are responses to this. The most common is that we are young, technologically, and that older civilisations would have moved beyond the listening phase into the broadcasting phase. This is plausible. It is also unfalsifiable, because we have no older civilisation to observe. Another response is that we are, in fact, broadcasting, our radio, television, and radar emissions leak into space continuously, and have been doing so for a century. This is true, but the power of these leaked signals at interstellar distances is far below the detection threshold of any plausible receiver. A civilisation like ours, at 1,000 light-years, would not be detectable by our own instruments. The leakage is real but too faint to be the signal we are searching for.
The third response is the one that the semiotic analysis suggests: that we are not broadcasting because broadcasting, as a strategy for interstellar communication, does not make sense. The energy costs are prohibitive. The probability of being heard is unknown. The temporal coincidence required is, on any reasonable estimate, small. And the assumption that another civilisation would be listening, in the same frequency band, with the same model of what a signal looks like, during the same century, is a tower of assumptions, each individually uncertain, that collectively approaches the kind of improbability that, in any other scientific context, would be treated as fantasy.
VI. The SETI Industry
SETI has existed, as a formal enterprise, since 1960, when Frank Drake conducted Project Ozma, the first systematic radio search for extraterrestrial signals, using the 26-metre radio telescope at Green Bank, West Virginia. Drake listened to two stars, Tau Ceti and Epsilon Eridani, for a total of approximately 150 hours, at a single frequency (1.42 GHz), and heard nothing. The search has grown since then, in telescope size, in frequency coverage, in target number, in computational sophistication. It has not, in sixty years, found anything.
This is not a failure, or not only a failure. It is a fact, and the fact has a context: the search space is vast, and the fraction of it we have explored is small. Jill Tarter, the longtime director of the Center for SETI Research, compared the explored volume of SETI’s search space to a glass of water taken from the ocean. If you take one glass of water from the ocean and find no fish, you have not proven that the ocean contains no fish. You have proven that this glass of water contains no fish. The ocean is large.
The metaphor is apt, and it is also, in its way, a defence of the enterprise against its own null results. The ocean is large. We have sampled a glass. The absence of fish in the glass is not evidence of the absence of fish in the ocean. This is true. It is also, in a practical sense, the kind of argument that can be used to justify indefinite searching, because the ocean is always larger than the glass, and the glass can always be refilled, and the fish can always be somewhere else.
The question is not whether the search should continue. The question is what the search costs, and what it produces, and whether the cost and the production are in a relationship that is sustainable or honest.
Funding
SETI has historically been funded by a mix of NASA, the National Science Foundation, and private sources. NASA’s funding for SETI was terminated by Congress in 1993, largely through the efforts of Senator Richard Bryan of Nevada, who declared that the search had found nothing and was a waste of taxpayer money. The field survived on private funding, the SETI Institute, founded in 1984, built the Allen Telescope Array with support from Paul Allen; the Breakthrough Listen initiative, launched in 2015 with $100 million from Yuri Milner, is the largest and best-funded SETI project in history, and it has access to some of the world’s most powerful radio telescopes, including the Green Bank Telescope in West Virginia and the Parkes Telescope in Australia.
One hundred million dollars is, by the standards of scientific funding, not a great deal of money. The James Webb Space Telescope cost approximately $10 billion. The Large Hadron Collider cost approximately $9 billion. Breakthrough Listen, spread over ten years, costs $10 million per year. In the context of global scientific expenditure, this is a rounding error.
But the comparison cuts both ways. If $100 million is a rounding error, then it is also $100 million that is not being spent on other things. The opportunity cost is not theoretical, it is the science that could have been done with the telescopes, the computing power, and the human capital that Breakthrough Listen has absorbed. The Green Bank Telescope, which spends a portion of its observing time on SETI, is a general-purpose radio telescope that could be used for pulsar timing, molecular spectroscopy, galaxy surveys, and the study of the interstellar medium. The time it spends listening for aliens is time it is not spending on those things. The computing infrastructure that processes SETI data, and SETI generates enormous volumes of data, requiring significant processing, could be processing data from other astronomical observations.
The question is whether the expected return on SETI justifies the opportunity cost. The expected return is the probability of a detection multiplied by the value of a detection. The value of a detection is, by any measure, immense, a confirmed signal from an extraterrestrial intelligence would be the most significant scientific discovery in human history, and its implications would extend far beyond science into philosophy, religion, politics, and culture. But the probability is, as we have established, not a number we can estimate. It is, at best, a range so wide that the expectation value is not well-defined. If the probability is 10⁻²⁰, the expected return is negligible. If the probability is 10⁻², the expected return is enormous. We do not know where, in that range, the truth lies, and the range is so wide that multiplying it by any value produces a result that can be used to justify either continuing or stopping, depending on which end of the range you prefer.
This is not a scientific question. It is a question about the allocation of resources in the face of radical uncertainty, and it is a question that the SETI industry, as an institution, has a structural incentive to answer in one direction.
The Industry
SETI is, at this point, an industry. It has institutions (the SETI Institute, the Berkeley SETI Research Center, the International Centre for Radio Astronomy Research). It has a workforce, scientists, engineers, software developers, administrators. It has a public profile, maintained through documentaries, popular books, conference talks, and the occasional viral news story about a “candidate signal” that turns out to be terrestrial interference or a known astrophysical phenomenon. It has a narrative, and the narrative is compelling: we are searching, the search is hard, the universe is vast, and the answer, if it comes, will change everything.
Industries do not, as a rule, argue for their own dissolution. The SETI industry is no exception. The null results of the last sixty years have been interpreted not as evidence to revise the fundamental assumptions of the search but as evidence to expand it: more telescopes, more frequencies, more stars, more sensitivity, more computing power. The logic is that the search space is large and we have explored a small fraction of it, and therefore the appropriate response to null results is to search more, not to question whether the thing we are searching for exists in the form we are searching for.
This is not unreasonable. It is also not science, in the strict sense, because the core hypothesis, that a technological civilisation would produce a signal recognisable to us, is not falsifiable by the search. If we search the entire sky, at all frequencies, for a century, and find nothing, the response can always be: the signal is there, but we are not looking at the right time, or the right modulation, or the right encoding, or with the right model. The hypothesis adapts to survive the evidence, and a hypothesis that adapts to survive all evidence is not a scientific hypothesis. It is an article of faith.
There is a comparison to be made with particle physics. The search for the Higgs boson was, for decades, a search for a predicted but unobserved phenomenon. The difference is that the Higgs was predicted by a theory, the Standard Model, that made other, testable predictions, and the search for the Higgs was constrained by those predictions. If the Higgs had not been found in the predicted mass range, the Standard Model would have been falsified, and the search would have had a defined endpoint. SETI has no equivalent. There is no theory that predicts, with specific parameters, what a signal from an alien civilisation would look like, at what frequency, with what modulation, from what direction. There are conjectures, the water hole, the hydrogen line, the beacon hypothesis, but they are not predictions in the scientific sense, because they are not derived from a testable theory. They are arguments from analogy and from aesthetics, and arguments from analogy and aesthetics, however appealing, are not falsifiable.
The Effect on Science
The more subtle cost of SETI is not the money or the telescope time. It is the effect on the scientific culture, specifically, on the culture of astrobiology, the broader field that studies the origin, evolution, and distribution of life in the universe.
Astrobiology is a legitimate and thriving science. It studies the chemistry of life’s origins, the conditions under which life can arise, the limits of life as we know it (extremophiles, subsurface biospheres, alternative biochemistries), and the detectability of life on other planets through biosignatures, atmospheric, geological, spectral. It is a field constrained by data: we can study extremophiles in the lab, we can analyse the atmospheres of exoplanets with telescopes like JWST, we can model the conditions on Mars, Europa, Enceladus, Titan. The data are limited but real, and the hypotheses are testable.
SETI, as a subfield of astrobiology, has a tendency to dominate the public perception of the whole. When people think about the search for extraterrestrial life, they think about radio signals and intelligent aliens, not about methane plumes on Mars or phosphine on Venus or the spectral signature of vegetation on an exoplanet. This is a problem, because the more immediately testable and productive lines of astrobiological research, the search for biosignatures, the study of extremophiles, the characterisation of exoplanet atmospheres, are, in the public mind, subordinated to the more speculative and less productive search for engineered signals.
The effect is compounded by the media, which prefers the SETI narrative because it is dramatic: signals from the stars, first contact, the question of whether we are alone. The discovery of a biosignature on a distant exoplanet would be a profound scientific result, but it would not, in the public imagination, compete with a signal from an alien intelligence. The SETI industry, by maintaining its public profile, shapes the public’s understanding of what astrobiology is and what it should fund, and the shape it imposes is one that privileges the speculative over the empirical.
There is a further effect, less visible but more insidious: the SETI framing can distort the kinds of questions that scientists ask. The assumption that intelligence is the thing to search for, that intelligence is the endpoint of life, the thing that matters, can bias the kinds of biosignatures we look for and the kinds of planets we prioritise. We search for Earth-like planets because we assume that Earth-like planets are the most likely to produce Earth-like intelligence. This is reasonable, but it is also circular: we search for ourselves, and we justify the search by the assumption that the universe produces beings like us, and the assumption is grounded in the only data point we have, which is ourselves.
VII. What If the Signal Is Not a Signal?
Let us return to semiotics, because semiotics is the discipline that can most productively interrogate the assumptions of the search.
If an alien intelligence exists, and if it communicates, the communication need not take the form of a narrowband radio signal modulated with a pattern recognisable to human cryptanalysis. It could take a form that we do not recognise as communication, because the form is grounded in a semiotic world that does not overlap with ours.
Consider: a civilisation that communicates through gravitational wave modulation. This is theoretically possible, a sufficiently advanced civilisation could, in principle, manipulate massive objects to produce gravitational waves with a structured pattern. We have only recently developed the technology to detect gravitational waves at all (LIGO, 2015), and our sensitivity is limited to the most violent astrophysical events, merging black holes and neutron stars. A modulated gravitational-wave signal from an alien civilisation would be, with current technology, undetectable. But it is not impossible, and if it is happening, we would not know.
Consider: a civilisation that communicates through neutrino beams. Neutrinos interact so weakly with matter that they pass through planets undisturbed, which makes them, in principle, an excellent medium for interstellar communication, no absorption, no scattering, no interference. But the difficulty of generating and detecting neutrinos is, by current technology, prohibitive. We can detect neutrinos from the sun and from supernovae, but only with enormous detectors buried deep underground, and only at very low event rates. An engineered neutrino signal would be, with current technology, indistinguishable from background.
Consider: a civilisation that communicates through modifications to its environment that are detectable at interstellar distances, a Dyson structure that blocks or modulates the light of its star in a non-natural pattern. This is the technosignature approach: rather than searching for a signal, search for the artefact. A megastructure around a star would produce a distinctive light curve, a non-periodic, non-natural dimming pattern that could not be explained by planetary transits or stellar variability. The Kepler mission, which surveyed a portion of the sky for transiting planets, produced light curves for hundreds of thousands of stars, and a few, most notably Tabby’s Star (KIC 8462852), showed dimming patterns that were, briefly, consistent with a megastructure hypothesis. The dimming was later explained by dust, but the episode illustrates the principle: technosignatures are, in principle, detectable with existing technology, and they do not require the alien civilisation to be deliberately communicating. They require only that it is doing something big enough to see.
The technosignature approach is, in semiotic terms, a shift from searching for a sign that is intended as communication to searching for an index, a sign that bears a physical connection to its object, like smoke to fire. An index does not require intent. It requires only a causal relationship between the sign vehicle and the thing it signifies. A Dyson structure is an index of a technological civilisation, whether or not the civilisation wants to be seen.
This is, arguably, a more honest search, because it does not require us to assume that an alien intelligence shares our semiotic assumptions about communication. It requires only that an alien intelligence, if it exists and is sufficiently advanced, will do things that are detectable, will modify its environment in ways that are, to a sufficiently careful observer, non-natural. The search for technosignatures is, in this sense, a search for the side effects of intelligence rather than for intelligence itself. It is less ambitious, less romantic, and more grounded in observable fact.
VIII. The Honest Position
The honest position, given everything we know, is this:
We do not know whether there is other intelligent life in the universe. We have one data point, ourselves, and one data point cannot populate a distribution. We do not know whether intelligence is rare or common, whether it tends toward communication or toward silence, whether it produces signals we would recognise or signals we would not. We do not know whether the assumptions underlying SETI, that mathematics is a universal language, that the hydrogen line is a cosmic meeting place, that an alien intelligence would broadcast in a form we can detect, are correct or are projections of our own semiotic world onto the void.
What we know is that the universe is large, that the conditions for life exist in many places, and that life, once it arises, is persistent and adaptable. We know that the search for life, not necessarily intelligent life, but life in any form, is a scientific enterprise with testable hypotheses and available data. We can search for biosignatures in the atmospheres of exoplanets. We can study the chemistry of the interstellar medium. We can explore the moons of our own solar system for subsurface oceans and the chemical signatures of metabolism. We can do these things now, with existing technology, and the results, whether positive or negative, will be informative.
The search for extraterrestrial intelligence, in its current form, is a search conditioned on assumptions that are not testable, funded by resources that have opportunity costs, and sustained by an industry that has a structural incentive to continue regardless of results. It is not, in its current form, a science, because its core hypothesis is not falsifiable. It is a practice, a disciplined, methodical, technologically sophisticated practice, and the practice may, one day, produce a result. But the practice is not the same as the science, and the distinction matters, because the conflation of the two, the tendency to treat SETI as if it were as rigorously grounded as, say, the search for exoplanets or the study of cosmic microwave background anisotropies, distorts the allocation of scientific resources and the public understanding of what we know and what we do not.
The semiotic perspective offers a way out of this, not a solution, but a clarity. If we understand that the search is, inescapably, a search for ourselves, for a mind like ours, producing a signal like ours, in a form we can recognise, then we can be honest about what a null result means. It means that the specific kind of mind we are searching for has not been found in the specific places we have looked. It does not mean that minds do not exist. It does not mean that the universe is empty. It means that our model of what a mind looks like, when it communicates across the void, has not been confirmed, and that the model is derived from a sample of one and may be, in ways we cannot detect, wrong.
The most productive thing we can do, in the face of this uncertainty, is to broaden the search, to look not only for the signals we expect but for the indices we do not. To search for biosignatures as well as technosignatures. To search for the side effects of life as well as the deliberate productions of intelligence. To search, in short, not for a mirror but for a window, a way of seeing the universe that does not assume that what we see will look like us.
This is not a argument for stopping SETI. It is an argument for situating it, for understanding its place in the larger enterprise of astrobiology, for being honest about its assumptions, for acknowledging its opportunity costs, and for refusing the temptation, which all industries feel, to justify their own continuation by the infinity of the search space and the impossibility of proving a negative.
The universe may be full of minds. It may be empty of them. We do not know, and the not-knowing is the honest position, and the honest position is the one from which the best science is done, because the best science is done not by those who are certain of what they will find but by those who are honest about what they do not.
The signal, if it comes, will not be what we expect. It will be what it is. The question is whether we are listening in a way that allows us to hear it.
A true account, as near as memory and the surviving women allow.*
They still call it that in the pubs and the corner shops of the estate, even though the papers long ago stopped using the phrase. Comet Babies. 12 March 1994. A clear, cold sky over northern England, the kind that makes the sodium lights look sickly. Something bright and dirty streaked across the dark, broke up, and fell in pieces that nobody officially recorded. Most people slept through it. Some woke to a soft rain that left a faint metallic smell on the washing lines and the pavements. By morning the smell was gone. By the end of the week the first women were late.
What follows is pieced together from the women who lived it, from the few medical notes that weren’t shredded, and from the quiet, stubborn records kept on the estate itself. Names have not been changed. The women asked that they not be.
Karen
Karen Ward was thirty-two, two kids already, living in a three-bed on the top walk of the Houghton estate. The kind of place where the lifts broke every other month and the bins overflowed because the council only came when they felt like it. Her husband, Dave, drove wagons and was away more than he was home. She worked mornings at the Co-op checkout and evenings cleaning offices. She was tired in the ordinary way.
The first sign was the hunger.
It came on the third day after the comet, sudden and total. She woke at four in the morning with her stomach clenched so hard she thought she was going to be sick, but there was nothing to bring up. She ate the last of the bread, then the biscuits, then the cold leftovers of the kids’ fish fingers. Still not enough. By the time the shops opened she was light-headed. She bought two loaves, a block of cheese, a packet of digestives, and ate half of it walking home. The other half she finished sitting on the sofa while the kids got themselves ready for school.
She told herself it was stress. Or the flu. Or the fact that Dave had been gone ten days and the heating was playing up again.
That night the cramps started. Low, deep, like period pain but wrong—too early, too sharp. She took paracetamol and went to bed early. In the dark she felt something move inside her, a slow roll that made her press both hands flat to her belly. She had two children. She knew what a baby felt like when it turned. This was smaller. And colder.
By the end of the week she had missed her period. The home test was positive. She sat on the closed toilet lid staring at the two blue lines and felt nothing at first except a dull, practical dread. Another mouth. Another set of shoes. Dave would go mental.
She booked in at the surgery.
Dr Patel was kind in the way overworked GPs are kind—efficient, slightly distant. He confirmed the pregnancy, estimated six weeks, and wrote her a referral for the twelve-week scan. Karen left with a leaflet about folic acid and a sense that something was already wrong. The hunger had not eased. She was eating constantly now: toast, crisps, whole packets of biscuits, the children’s leftover chips. Her belly stayed flat. No swelling. Just the constant, grinding need for food and the occasional flutter deep inside that felt nothing like a human child.
Two other women on the estate were late as well. Kaylie from number 47, nineteen and already a mum to a toddler. Latoyah from the flats near the shops, twenty-four, no kids yet, lived with her mum. They started comparing notes in the corner shop while buying more bread than any one household needed.
“I’m starving all the time,” Kaylie said, voice low so the cashier wouldn’t hear. “Like, proper starving. And I’m never late. Never.”
Latoyah nodded. “Same. And I keep feeling it move. Tiny. Like a fish or summat.”
Karen said nothing at first. Then: “I took a test. It’s positive.”
The three of them stood there among the tinned beans and the lottery tickets and felt the first real cold of it settle in.
The news broke slowly. First a few more positive tests on the estate. Then a cluster reported at the local hospital. Then the regional papers picked it up: unexplained surge in early pregnancies across three northern towns, all dating to the same narrow window after the comet. The national press arrived a week later with their cameras and their headlines about “Comet Babies” and “Alien Conception Scare.” Scientists muttered about radiation or chemical fallout or mass hysteria. The Ministry of Defence said nothing useful. The NHS began calling women in for mandatory scans and bloods.
Karen went. She had no choice if she wanted the child benefit forms sorted.
The ultrasound room was colder than it should have been. The gel was cold. The probe pressed hard. The sonographer’s face changed.
“There’s… multiple sacs,” she said carefully. “But only one seems viable at the moment. The others are… irregular.”
Karen stared at the screen. A dark shape, small, coiled. Not a bean. Not a human outline. Something with too many thin, shifting lines.
“What is that?”
The sonographer didn’t answer. She called for a consultant. The consultant looked, then looked again, then asked Karen to wait outside while they made phone calls.
That afternoon an official in a grey suit and a woman from public health sat her down and told her the pregnancy was non-viable. Abnormal. A risk to her health. Termination was strongly advised—recommended—required, if she wanted continued care. They used soft words. Clinical words. They handed her a consent form already half-filled.
Karen signed because she was frightened and because they made it sound like the only sane thing to do. She went in the next morning. D&C. Local anaesthetic. They told her it would be over quickly.
It wasn’t.
Something in her resisted. The procedure took longer than it should have. She felt pulling, scraping, a deep wrongness. When they finished, the nurse looked pale. The specimen jar was taken away under a cloth.
Karen went home empty and sore and still hungry.
Three days later the hunger returned harder than before. The cramps came back. Another positive test. Another scan. Another “non-viable” diagnosis. Another termination.
By the fourth time the doctors stopped pretending it was ordinary. They admitted her. Put her on a drip. Started talking about “recurrent anomalous implantation” and “possible parasitic gestation.” They used the word “changeling” only once, in a corridor when they thought she couldn’t hear.
On the estate the other women were watching. Some went quietly to the hospital and never came back the same. Some disappeared into the system of wards and locked doors and repeated procedures. A few, like Karen, kept coming home between the terminations, thinner, greyer, still eating everything in the house.
Kaylie stopped answering her door.
Latoyah’s mum started telling people her daughter had gone to stay with relatives down south.
Chris—Christopher Hale, twenty-eight, father of Kaylie’s toddler—started asking questions at the pub that nobody wanted to answer.
Ivy, who ran the community centre and knew every family on the estate by name, began keeping a quiet list of who was late, who had been called in, who had come back empty-eyed.
Susanna, the health visitor who still made her rounds despite the official advice to report every anomalous case, started leaving extra food parcels on certain doorsteps and saying nothing about it in her notes.
The hunger did not stop. The things inside the women did not stop growing back.
And on the Houghton estate, in the damp flats and the cracked walkways under the orange lights, the women began to choose sides without ever saying the words out loud.
Kaylie
Kaylie Brennan was nineteen and already tired of being looked at like she was thick. She lived in number 47 with her two-year-old, Jamie, and the occasional presence of Chris, who was the dad and sometimes slept on the sofa and sometimes disappeared for days when the money ran out or the arguments got too sharp. The flat smelled of damp and washing powder and the cheap air freshener she sprayed to cover both. She kept it clean because that was one of the few things she could control.
The hunger hit her harder than it hit most.
It started the morning after the comet. She woke with her mouth dry and her stomach already growling, even though she’d eaten a proper tea the night before—chips, beans, a bit of sausage for Jamie. She made toast. Ate four slices. Still empty. By midday she was going through the cupboards like someone had robbed her. Jamie watched from his highchair, solemn, while she finished the last of the cereal dry because the milk had gone.
“Mam’s just a bit poorly, love,” she told him, though she didn’t feel poorly. She felt hollowed out and ravenous at the same time.
Her period was due that week. It didn’t come. She’d never been regular exactly, but she’d never been this late either, not since Jamie. She waited another week, then bought a test from the chemist on the high street, the expensive one because she didn’t trust the cheap ones. Two lines. Clear as anything.
She sat on the edge of the bath with the stick in her hand and cried without making a sound so Jamie wouldn’t hear. Another baby. Chris would do his nut. The council would want to know about the bedroom tax and the benefits. She was already skint.
The movement started a few days later. Not kicks. Something smaller, wetter, sliding under the skin low down. She pressed her fingers there and felt it shift away, like it didn’t want to be touched. That night she dreamed of cold water and something coiled in the dark.
She told Chris when he next turned up, two days later, smelling of the pub and looking sheepish.
“I’m pregnant.”
He stared at her. “You what?”
“Test said so. And I can feel it. Something’s not right, Chris. It’s moving already and it’s only been a couple of weeks.”
He rubbed his face. “Fuck’s sake, Kaylie. We can’t afford another one. Have you been down the doctor’s?”
“Not yet.”
“Well go. Get it sorted. I’m not having another kid on this estate with nowt coming in.”
She went. Same surgery as Karen. Same Dr Patel, who looked more tired than last time and didn’t meet her eyes properly when the test confirmed it. He booked the scan and gave her the same leaflets. She didn’t tell him about the hunger or the cold sliding feeling. She didn’t know how to put it into words that wouldn’t make her sound mad.
The scan was worse than she expected.
The room was full of machines and the woman doing it kept glancing at the door like she wanted someone else to come in. On the screen there was a dark shape, not the little bean Kaylie remembered from Jamie. This one was longer, thinner, with pale threads that twitched when the probe moved. The sonographer went very quiet. Then she said, carefully, “I’m going to need to get a second opinion.”
They left Kaylie on the couch with the gel drying sticky on her belly. She heard low voices in the corridor. When they came back it was a different doctor, older, and a woman in a suit who wasn’t wearing a white coat.
They told her the pregnancy was abnormal. Non-viable. A danger. Termination was the only safe option. They used the word “urgent.” They had a slot the next morning.
Kaylie signed because Chris had said get it sorted, and because the way they looked at her made her feel like a child again, thick and in the way. She went home and cried into Jamie’s hair while he watched cartoons. Chris didn’t come back that night.
The procedure was supposed to be straightforward. It wasn’t. She felt them scraping and the thing inside her twisting away, and for a moment she thought she heard a thin wet sound that wasn’t the machines. When they finished she was shaking. They sent her home with painkillers and a leaflet about counselling.
Three days later the hunger came back like a punch. The test was positive again. The movement returned, lower this time, more insistent.
She didn’t go back to the surgery straight away. She stayed in the flat with the curtains drawn and ate everything she could get her hands on—bread, biscuits, the emergency tins at the back of the cupboard, even the dry pasta when there was nothing left. Jamie started crying because there was no food left for him. She walked to the shop with him in the pushchair and spent the last of the child benefit on more bread and milk and a big bag of potatoes. The woman behind the counter looked at the pile and then at Kaylie’s face and didn’t say anything.
On the way home she saw Karen coming out of the lifts, thinner than she’d been a fortnight ago, eyes shadowed.
“You as well?” Karen asked quietly.
Kaylie nodded. “They took it out. It’s back.”
Karen’s mouth tightened. “Same. Third time for me. They’re talking about keeping me in next time.”
They stood there in the wind that always funneled down the walkway, two women with empty bellies that wouldn’t stay empty, while Jamie kicked his feet against the pushchair and asked for a biscuit.
That was the day Kaylie stopped answering the phone when the surgery called. She stopped opening the door when the health visitor knocked. She started keeping a bag of clothes and nappies by the door in case she needed to leave quick. She didn’t know where she would go. She only knew she wasn’t going back under those lights and those scraping instruments if she could help it.
Chris turned up again a few nights later. She told him the truth this time—the whole of it, the repeated terminations, the thing growing back, the way the doctors looked at her like she was already a problem to be managed.
He went pale. Then he got angry, the way he did when he was frightened.
“You’re talking shite. They’re doctors. If they say it’s wrong, it’s wrong. You need to go back and get it done proper.”
“I can’t,” she said. “It doesn’t stay gone. And I’m not letting them cut me open again while I’m awake. I felt it, Chris. It didn’t want to come out.”
He left. Slammed the door hard enough to make Jamie cry. She didn’t see him for a week after that.
In the quiet that followed, Kaylie began to notice the other women. Latoyah from the flats, who had stopped coming to the community centre. Ivy, who left a bag of shopping outside number 47 one morning with a note that just said Eat what you need. Tell no one. Susanna, the health visitor, who still knocked but when Kaylie finally opened the door a crack only said, “I’m not here officially. How many weeks do you think you are this time?” and then left without writing anything down.
The hunger grew. The thing inside her grew with it. She could feel the shape of it now when she lay still—long, coiled, the faint brush of something like thin limbs or tentacles against the wall of her womb. It didn’t kick like Jamie had. It pulsed. And every time it pulsed she felt a wash of something that wasn’t quite love and wasn’t quite fear, a biological tug that made her press her hands over it protectively even while her mind screamed that it was wrong.
She started talking to it in the dark, the way she had with Jamie when she was pregnant the first time.
“You’re not supposed to be here,” she whispered. “But you’re here now. So we have to figure something out.”
It didn’t answer. It only shifted, cold and patient, and the hunger deepened.
By the end of the month the papers were full of it. Clusters of anomalous pregnancies. Mandatory reporting. Hospitals opening special wards. Women being sectioned under mental health acts when they refused treatment. On the Houghton estate the women who were still free began to move differently—quieter, more watchful, sharing food and information in the corners of the community centre when Ivy looked the other way.
Kaylie kept her curtains closed. She fed Jamie first, then herself, then the thing that was growing back for the third time. She waited. She did not know for what. Only that she would not go back to the hospital if she could help it, and that the cold sliding presence inside her was starting to feel less like an invader and more like something that belonged to her, whether she wanted it or not.
Latoyah
Latoyah Okonkwo was twenty-four and still lived with her mum in the low-rise flats near the shops because the waiting list for a place of her own was years long and the private rents were a joke. She worked nights at the care home on the edge of town, wiping arses and making tea for old people who sometimes remembered her name and sometimes called her by the names of daughters long gone. She was good at the job. Quiet, steady, didn’t flinch. Her mum, Grace, cleaned offices in the city centre and came home smelling of polish and exhaustion. They got on. They kept out of each other’s way when they needed to.
The hunger arrived like a betrayal.
Latoyah had always been careful with food. Not fussy—just measured. She knew what it cost. After the comet she woke one morning with her stomach clenched so tight she had to sit on the edge of the bed until the dizziness passed. She ate the leftover rice from the night before, then the bread, then the last of the groundnuts her auntie had sent from London. Still not enough. By the end of the shift at the care home she was light-headed and snapping at the residents, which wasn’t like her.
She missed her period the following week. She had never been late. Not once. She bought a test on the way home, did it in the staff toilet at the care home because she didn’t want her mum to see the packet, and stared at the two lines until the timer ran out and the result stayed the same.
She told no one at first.
The movement started early. Too early. A slow, cold coiling low in her belly that made her press her palm there during the night feeds at work, pretending she was just tired. It didn’t feel like a baby. She had helped deliver enough of them in the care home when the old women still talked about their own labours. This was different. Slippery. Insistent. Like something trying to find the right place to settle.
She went to the surgery because the hunger was making her drop things at work and the manager had started asking questions. Dr Patel—same one who had seen Karen and Kaylie—confirmed the pregnancy and booked the scan with the same careful, distant voice. Latoyah noticed the dark circles under his eyes. She noticed the way the receptionist looked at her when she gave her name, like another tick on a list that was getting too long.
The scan showed what the others had shown. Multiple irregular sacs. One larger shape that moved when the probe pressed, thin pale structures shifting inside the dark. The sonographer went quiet. The consultant came in. Then the woman from public health who wasn’t wearing a badge that Latoyah recognised.
They used the same words. Non-viable. Abnormal. Risk to maternal health. Termination recommended—required if she wanted to remain under NHS care. They had a slot the next day. They spoke gently. They watched her face the way people watch someone who might make a scene.
Latoyah signed. She was practical. She had seen what happened to women who argued with systems. She went home, told her mum she had a stomach bug, and went in the next morning.
The procedure took longer than they said it would. She felt the scrape and the pull and a sudden deep resistance that made the doctor swear under his breath. Something came out that they covered quickly. She was sent home with painkillers and instructions to rest. Her mum made soup. Latoyah ate it and then the rest of the loaf and still felt hollow.
Four days later the hunger returned. The test was positive again. The cold coiling feeling was back, lower and stronger.
She did not go back to the surgery.
Instead she started moving differently. She swapped her night shifts for days so she could be home when the health visitor might call. She stopped answering the landline. She began buying food in cash from the corner shop and the market stalls, more than two women needed, hiding the excess under her bed in the way she had once hidden exam revision from her mum when she was a teenager. She ate in the dark after Grace had gone to sleep—bread, tins of beans, packets of biscuits, cold rice, anything that filled the space for an hour or two.
The thing inside her grew. She could map its shape now when she lay still. Long. Segmented. The faint brush of what felt like thin, flexible limbs against the inner wall. Sometimes it pulsed and a wave of heat went through her that wasn’t fever. Sometimes it went completely still and the hunger sharpened until she had to get up and eat or risk being sick from the emptiness.
She started talking to Karen in the stairwell when they crossed paths, voices low.
“How many times for you?” Latoyah asked.
“Four. They’re keeping me in next time. Said the regeneration is stressing the uterus. Whatever that means.”
“I’m not going back.”
Karen looked at her for a long moment. “Neither is Kaylie. She’s stopped answering the door. Ivy’s been leaving food.”
Latoyah nodded. She had seen the bags outside certain flats. She had seen Susanna the health visitor walking the estate with her usual bag but not writing in her notebook the way she used to.
The papers had caught up by then. “Comet Pregnancy Crisis.” “NHS Overwhelmed by Anomalous Gestations.” “Experts Warn of Unknown Pathogen.” Television crews tried to film on the estate and were told to clear off by groups of lads who didn’t like cameras pointed at their mams and sisters. The official line was that the pregnancies were the result of an unidentified environmental contaminant linked to the meteor event. Termination and supportive care were the only recommended pathways. Women who refused were to be reported. Some were already being held on the special wards at the general hospital—drips, monitors, repeated procedures as the things grew back one after another. Word filtered back through cousins and friends who worked there: the women on the wards were weakening. The repeated growth and removal was taking too much. Some had started haemorrhaging. Some had simply stopped fighting the next implantation.
On the Houghton estate the ones who were still outside began to organise without ever calling it that.
Ivy opened the community centre earlier and locked the doors when the wrong people came asking questions. She kept a quiet tally of who was pregnant, who had been taken in, who was still free. She made sure there was always tea and biscuits and a back room where women could sit without being overheard.
Susanna stopped filing the full reports. She still visited. She still checked blood pressures and asked about bleeding. But she left extra iron tablets and bags of rice and never wrote down the true dates or the true symptoms.
Chris, Kaylie’s on-and-off, started turning up at the pub with questions that made the older men look at their pints. He had seen the scan pictures Kaylie had stolen from her notes before the second termination. He had seen the shape. He didn’t know what to do with the knowledge, so he drank and asked the same questions of anyone who would listen.
Latoyah kept her head down and her curtains closed and the thing inside her fed. She could feel it more clearly every day—the slow stretch of it, the way it seemed to settle deeper when she ate, the strange protective heat that rose in her chest when she thought about the hospital and the scraping instruments. Biological imperative, the doctors had called it in one of the leaflets she had been given and then thrown away. She didn’t have a better word. She only knew that the thought of letting them take it again made her hands shake and her vision narrow.
One night her mum came into her room without knocking. Grace stood in the doorway in her dressing gown and looked at the empty plates stacked by the bed and at Latoyah’s hands resting over her still-flat belly.
“How far along are you this time?” Grace asked quietly.
Latoyah didn’t lie. “Maybe seven weeks. Maybe more. It doesn’t grow like a normal one.”
Grace was silent for a long time. Then she said, “Your auntie in London says they’re taking women off the streets down there if they refuse the procedure. Putting them on wards and not letting family in. I’m not having that. Not with you.”
She sat on the edge of the bed. “We hide it. We say nothing. When the time comes we do it here. Ivy knows people who can help with the birth if it comes to that. Home birth. Quiet. No hospitals.”
Latoyah felt the thing inside her shift, as if it had heard. The hunger surged. She reached for the packet of biscuits on the side and ate three before she could answer.
“They’re not human, Mum.”
“I know.” Grace’s voice was steady. “But they’re in you. And you’re mine. So we deal with what’s in front of us.”
That was the night Latoyah stopped thinking of escape as a solitary act. She was one of several now. Karen on the top walk, still going in and out of the hospital but talking more to the others between admissions. Kaylie locked in with Jamie and the growing cold presence she had started calling “it” out loud. Ivy keeping the lists and the back room and the extra food. Susanna walking the estate with her mouth shut and her bag heavier than it should have been. Chris asking questions that no one in authority wanted to answer.
And somewhere in the general hospital the women who had not been able to refuse were hooked to drips and monitors while the seeds grew back one after another, each removal taking a little more of them, each new implantation arriving faster than the last.
Latoyah lay awake after her mum had gone and felt the slow pulse of the thing that would not stay gone. She did not love it. Not yet. But the thought of the hospital lights and the instruments made her curl around her belly protectively, and the thing inside her seemed to settle, as if it understood the bargain that was forming in the dark.
Chris
Chris Hale was twenty-eight and had never been good at staying. He worked cash-in-hand on building sites when there was work, signed on when there wasn’t, and spent too many nights in the Crown & Anchor because the flat with Kaylie and Jamie felt too small and too full of things he didn’t know how to fix. He loved them in the way that made him angry when he thought about it too long. He had a temper that flared when he was frightened, and he was frightened a lot that spring.
He first heard about the comet the way most people did—someone in the pub saying there’d been a bright streak and a funny smell in the rain. He didn’t think anything of it. Then Kaylie told him she was pregnant and he did what he always did: shut down, got loud, told her to get it sorted. He hadn’t looked at her face properly when he said it. He had only thought about the money and the space and the fact that he already felt like he was failing at the one kid they had.
When she told him the second time—after the termination, after the thing grew back—he had called her mad and left. He stayed away a week. Slept on mates’ sofas. Drank. Tried not to think about the way her hands had rested over her belly while she spoke, protective even while her voice shook.
He went back because Jamie had started asking for him at the community centre, and Ivy had given him a look that said she knew exactly where he’d been hiding.
Kaylie opened the door with the chain on. She looked thinner in the face and heavier around the middle, though the pregnancy still didn’t show the way a normal one would. The flat smelled of toast and the cheap cleaner she used on everything.
“You coming in or what?” she said.
He came in. Jamie launched himself at his legs. Chris picked him up and felt the familiar mix of love and panic. Kaylie watched them without smiling.
“It’s still there,” she said once Jamie was occupied with a cartoon. “Growing. I can feel it proper now. Long. Like it’s got arms or tentacles or summat. And I’m hungry all the time. Proper starving. I eat and it’s like it takes it straight off me.”
Chris sat at the small kitchen table and stared at the laminate. “The doctors said it’s not right. They said get rid.”
“They did get rid. Twice. It comes back. Karen’s on her fifth. They’re keeping her in the hospital now. Latoyah’s not going back. Neither am I.”
He looked up. “You can’t just hide a pregnancy, Kaylie. They’ll find out. Benefits. Health visitor. School when Jamie starts. They’ll take him off you if they think you’re not coping.”
“They’re already taking women,” she said. “Putting them on wards and scraping them out over and over till there’s nowt left of them. I’m not going in there. Ivy’s helping. Susanna’s not reporting us. We’re doing it quiet.”
Chris didn’t know what to do with the information. He went to the pub that night and asked questions he shouldn’t have. Old Tommy behind the bar had a niece on one of the special wards. He said the women there were hooked up to drips because the repeated growths were draining them. Said some of them had started bleeding and not stopping. Said the doctors looked scared and the security had been doubled.
“They’re calling them changelings in the papers,” Tommy said, wiping a glass. “Load of shite. But whatever’s in them lasses isn’t coming out easy, and it’s not staying gone.”
Chris drank until the fear softened into something he could carry. The next day he went looking for answers in the only places he knew—mates who worked at the hospital, a cousin who drove ambulances, the quiet conversations at the bookies where men talked about their partners and sisters and the way the hunger made them eat everything in the house and still look empty.
He saw one of the scan pictures Kaylie had kept. She had stolen it from the file before the second procedure. The shape on the black-and-white print was wrong. Too elongated. Faint radiating lines that looked like thin limbs or feelers. A darker central mass that the radiographer had circled and then crossed out.
He felt sick. Then he felt angry. Then he felt the same protective heat he sometimes felt when Jamie was poorly—an animal thing that didn’t care about sense.
He started carrying shopping to number 47. Bread, milk, tins, the big bags of porridge oats that filled a stomach for longer. He fixed the broken catch on the kitchen window so Kaylie could leave it open at night without worrying about who might climb in. He sat with Jamie so she could sleep in the afternoons when the hunger and the movement made rest impossible.
He still didn’t know how to talk about the thing inside her. He called it “it” the way she did. He watched her hands rest over her belly and felt something twist in his own chest that wasn’t jealousy and wasn’t acceptance. It was closer to grief.
One evening he found her standing in the bathroom looking at herself sideways in the mirror. The bump was still small, lower than a normal pregnancy, and when she lifted her top he saw the faint movement under the skin—slow, coiling, like something turning over in cold water.
“Does it hurt?” he asked.
“No. It just… wants. All the time. Food. And me not to go near the hospital. I can feel it when I think about the wards. It goes still and cold and I start shaking.”
He stepped closer. She didn’t pull away. He put his hand over hers on her belly and felt the shift of it under the skin, a deliberate press against his palm as if it already knew the difference between threat and not-threat.
“I’m not going anywhere,” he said, and for once he thought he might mean it.
Outside the flat the estate was changing. More women with closed curtains. More quiet deliveries of food after dark. Ivy’s community centre stayed open later and the back room was always occupied. Susanna walked her rounds with a heavier bag and a closed mouth. Karen was still in the hospital; word came through a cleaner who lived on the next walk that she was on her sixth regeneration and the doctors were talking about hysterectomy if the next removal didn’t hold. Latoyah and her mum had stopped answering the door to anyone in a suit or a uniform.
The authorities had started knocking harder. Official letters. Visits from social services framed as welfare checks. Two women from the lower flats had been taken in the previous week after refusing scans; their kids had gone to temporary foster care and the mothers had not come back. The message was clear. Compliance or consequences.
Chris began walking Kaylie to the corner shop and back. He stood outside when the health visitor came and made sure it was Susanna and not one of the new ones who asked too many questions. He learned the names of the other women who were hiding—Karen’s neighbour who had gone to ground, a quiet girl from the tower block, two sisters on the far side of the estate who only came out at night. He didn’t know what he was becoming. He only knew that the alternative was watching Kaylie get taken the way the others had, and something in him refused that picture.
Late one night, after Jamie was asleep and Kaylie had finally dozed off with her hands still curved around the low swell of her belly, Chris sat in the kitchen with a mug of tea gone cold and tried to name what he was feeling. It wasn’t love for the thing inside her. He didn’t think he would ever manage that. It was something simpler and harder. She was his, in the messy, unfinished way of people who had made a child together and never quite sorted the rest. The thing inside her was part of her now, whether it should have been or not. And the people with the clipboards and the wards and the repeated procedures were coming for both of them.
He finished the tea. Rinsed the mug. Checked the locks. Then he lay down on the sofa the way he used to when they were first arguing about everything, and listened to the quiet of the flat, and the quieter sound of Kaylie breathing in the next room, and the knowledge that none of them were getting out of this clean.
Ivy
Ivy Morrison was forty-seven and had run the Houghton Community Centre for eleven years. She knew every family on the estate by the sound of their kids’ feet on the stairs and the particular way they argued when the benefits were late. She had buried two husbands, raised three children who had all left for better places, and kept the centre open through funding cuts, broken boilers, and the slow grinding knowledge that the council would rather the whole place quietly failed. She was wide, strong, grey at the roots, and had a voice that could stop a fight at twenty paces. People trusted her because she never pretended things were better than they were.
She first noticed the pattern in the second week after the comet.
Women who usually came in for the parent-and-toddler group or the cheap toastie lunch started missing sessions. When they did appear they were pale, distracted, always eating. The biscuit tin emptied faster than usual. The kitchen staff reported missing loaves. Ivy started watching more carefully. She saw Karen Ward leaving the centre one afternoon with two full carrier bags and a look that said she was already calculating how long the food would last. She saw Kaylie Brennan sitting in the corner of the main hall with Jamie on her knee, one hand pressed low on her belly while the other fed the boy bits of apple she wasn’t eating herself. She saw Latoyah Okonkwo collect a food parcel and then ask, very quietly, whether the back room could be used for a private conversation.
Ivy unlocked the back room and listened.
By the end of the month she had a list. Not written down—she wasn’t stupid—but held in her head the way she held the names of every child who used the after-school club. Twelve women on the estate with confirmed or suspected anomalous pregnancies. Four already taken into the general hospital’s new isolation wing. Three more receiving official letters that used words like “mandatory assessment” and “public health requirement.” The rest were still outside, hungry, frightened, and starting to look to her for something she wasn’t sure she could give.
She began leaving bags of shopping outside certain doors before the centre opened. Bread, oats, tinned fruit, UHT milk, the high-calorie stuff that would keep a body going when it was feeding something that didn’t belong there. She did it without notes and without being seen. When Susanna the health visitor came in for her usual cup of tea, Ivy looked at her over the rim of her mug and said, “You’re not writing everything down anymore.”
Susanna held her gaze. “Neither are you, from what I hear.”
They didn’t say anything else. They didn’t need to. The agreement settled between them like a third person in the room.
The centre became the quiet heart of it. Ivy opened earlier and closed later. She kept the back room free. Women started using it the way they had once used the old laundry rooms—places to talk without being overheard. They compared symptoms in low voices: the hunger that never eased, the cold coiling movement, the way the things inside them seemed to react to the mention of hospital, going still and heavy as if bracing. They shared what little they knew about the wards. Karen managed to get a message out through a cleaner she knew: the women inside were on constant drips. The regenerations were coming faster. Some of the mothers were developing fevers that didn’t respond to antibiotics. One woman from the tower block had haemorrhaged during a removal and been taken to intensive care. She hadn’t come back to the ward.
Ivy listened to all of it and kept her face steady. She had seen enough over the years to know that panic helped no one. What helped was food, locked doors, and a plan for when the births started.
Because they would start. The things were growing. Not at the rate of human babies—faster in some ways, slower in others—but they were growing. Kaylie was beginning to show properly now, a low, tight swell that moved under her clothes when she thought no one was looking. Latoyah had the same. Others were further along, hiding under baggy jumpers and the general tiredness of the estate that made people look away.
Ivy started talking to the older women who still remembered home births from before everything became hospital-default. She found two who were willing to help if it came to it—quiet, practical women who had caught babies in front rooms when the ambulances were too slow or the mothers too stubborn. She stocked the back room with clean towels, plastic sheeting, sanitary pads, a couple of birth pools that had been donated years ago for a water-birth group that never quite took off. She didn’t advertise what she was doing. She simply made sure the things were there.
Chris Hale started appearing more often. He carried shopping without being asked. He fixed the broken lock on the centre’s side door. He stood outside when official-looking cars pulled up and made sure the women inside had time to move to the back if needed. Ivy watched him and saw the same shift she was seeing in some of the mothers—the slow, reluctant acceptance that this was no longer a problem that could be handed over to someone in a white coat.
One afternoon Karen’s sister came in. She had been to the hospital. She sat in the back room with her hands wrapped around a mug of tea that Ivy had made extra strong and said, “They’re dying in there. Not quick. The things keep growing back and every time they take one the women get weaker. They’re talking about experimental treatments and consent forms that don’t mean anything anymore. Karen told me to tell you she’s trying to hold on but she doesn’t know how many more she can take.”
Ivy felt the familiar cold settle in her chest—the one that came when a child on the estate was being failed by every system that was supposed to protect them. She nodded once.
“Tell her we’re here. Tell her if she can get out, we have a place.”
She knew it was almost certainly too late for the ones already inside. The security on the isolation wing had been increased. Visitors were restricted. The official line was still contamination and non-viable gestation and the necessity of repeated intervention. The papers had moved on to other scandals, the way they always did, leaving the women on the wards as a quiet, ongoing failure that no one in power wanted to look at too closely.
On the estate the opposite was happening. The women who remained free were becoming more organised, more careful, more tightly bound to each other. They shared food. They watched each other’s kids. They developed signals—curtains left half-drawn, a particular plant pot moved to the left of a door—to show who was safe and who was under pressure. Latoyah’s mum, Grace, started coming to the centre with extra portions of food from her own kitchen. Kaylie brought Jamie and let him play with the toys while she sat with her hands over the moving shape inside her and listened to the others talk about what the births might look like.
No one pretended it would be ordinary.
Ivy had seen the scan images some of the women had managed to keep. She had heard the descriptions: tentacle-ringed mouths, dark eyes, bodies that were not human and not animal but something that had used the womb as a doorway. She had also seen the way the mothers already curled around the pregnancies when they felt threatened. Biological imperative. Pheromones. Whatever name the doctors wanted to give it. The bond was forming whether the women wanted it or not.
She didn’t know what would happen when the things were born. She only knew that the alternative—letting the hospital take every last one of them and scrape them out until the mothers were empty shells—was not a thing she could stand by and watch.
So she kept the centre open. She kept the back room ready. She kept the list in her head and the food moving and the doors locked against the wrong kind of visitors. And when the first of the free women went into early labour on a wet Tuesday night in late May, it was Ivy who got the call, and Ivy who unlocked the side door, and Ivy who said, calm as anything, “Right. We’re doing this here. No hospitals. No arguments. Let’s get her comfortable and see what we’re dealing with.”
Susanna
Susanna Reed was thirty-nine and had been a health visitor on the Houghton estate for six years. She knew the smell of the stairwells, the particular damp that never left the low-rise flats, the way mothers watched her bag when she opened it as if it might contain bad news. She was good at her job in the ordinary sense—patient, thorough, not easily shocked. She had seen neglect, violence, addiction, the slow grind of poverty that made every small crisis worse. She had never seen anything like the spring of 1994.
The first anomalous pregnancy she logged was Karen Ward’s. She had written the notes the way she always did: clear, dated, careful with the language. Positive test. Early scan anomalies. Referral to secondary care. When the termination was recorded and then the second positive test arrived within days, she wrote that down too. By the third regeneration she had started leaving things out of the official file.
It wasn’t a sudden decision. It was a series of small ones. The look on Kaylie Brennan’s face when Susanna asked about bleeding and the girl simply shook her head and pressed both hands over the low swell of her belly. The way Latoyah Okonkwo’s blood pressure stayed normal while everything else about the pregnancy was wrong. The quiet message that came back from the hospital through a nurse Susanna had trained with: the women on the isolation wing were not recovering between procedures. The seeds—someone had started using that word in the clinical notes—regrew too quickly. Nutritional demand was extreme. Several mothers were now on continuous feeding tubes because they could no longer eat enough by mouth to keep up. Two had developed sepsis after incomplete removals. One had died on the table during the fifth termination; the official cause was listed as haemorrhage secondary to abnormal placentation.
Susanna stopped writing the full truth the week that death was quietly confirmed.
She still made her visits. She still checked blood pressures and urine and the height of the fundus, though the measurements never matched any chart she had been trained on. She still asked the required questions. But she stopped recording the true dates. She stopped flagging the refusals. She began carrying extra high-calorie supplements in her bag and leaving them on kitchen tables without writing a prescription. When the new public-health directives came down—mandatory reporting of any suspected anomalous gestation, escalation of non-compliant cases to social services and, if necessary, mental-health assessment for capacity—she read them, signed the acknowledgement form, and then did the opposite.
She worked with Ivy without ever discussing the arrangement out loud. A look across the community-centre kitchen. A quieter-than-usual handover when a woman was close to crisis. The understanding that certain names were no longer being entered into the system the way the guidelines demanded.
The hospital team noticed. Of course they did. A consultant called her in for a “supportive discussion” about protocol and the importance of consistent data in an evolving public-health event. Susanna sat in the overheated office and listened to the careful, bureaucratic language and thought about Karen Ward, who was by then on her seventh regeneration and too weak to sit up unsupported. She thought about the scan images she had seen—the elongated shapes, the radiating filaments, the dark central masses that pulsed with their own rhythm. She thought about the mothers on the estate who were still free, still hungry, still curling around their secret pregnancies with a ferocity that looked like love even while their faces showed pure fear.
“I’ll take it under advisement,” she told the consultant, and left with her bag heavier than when she arrived.
On the estate the births were getting closer. Kaylie was the furthest along of the ones still outside. The bump was low and tight and moved constantly now, long slow rolls that made her breath catch. She had started leaking a thin, cloudy fluid that wasn’t milk and wasn’t normal show. Latoyah was not far behind. Two other women on the far side of the estate had gone into early labour already; Ivy had managed both in the back room of the centre with the help of the older women who still remembered how to catch a baby without a hospital around them. The first birth had taken six hours. The second, shorter. Both mothers had survived. Both infants—if that was the right word—had been alive.
Susanna had not been present for either. She had only seen the aftermath: the mothers pale and shaking and already protective, the small wrapped shapes that did not cry like human newborns. She had been told enough. Tentacle-ringed mouths. Dark, milky eyes. Skin that was mottled and slightly translucent in places. They had latched onto the breast with a strength that surprised everyone in the room. They took human milk. They settled when held. The mothers, despite everything, did not reject them.
Biological imperative. The phrase kept appearing in the restricted clinical bulletins Susanna still received. Whatever the organisms were, they exerted a powerful bonding effect on the host. Resistance was possible but costly. Most of the women on the wards who had undergone repeated terminations showed signs of profound psychological and physiological stress; the few who had been allowed to progress further (under heavy monitoring, in the name of research) showed the opposite—calm, focused, fiercely attached.
Susanna did not know which outcome was worse.
She kept walking her rounds. She kept leaving the extra food. She kept her mouth shut when the social-services teams came asking for lists of non-compliant mothers. When they pressed, she gave them the names of women who had already been taken or who had moved away. She protected the ones who were still hiding.
One evening she sat in her car outside the community centre after locking the drug cupboard and updating the incomplete files she still had to maintain. Rain streaked the windscreen. The orange sodium lights made the wet tarmac look like polished blood. She thought about the career she was quietly destroying, the professional standards she was choosing to breach every day, the possibility that she would be struck off if the full extent of her omissions ever came to light. Then she thought about Kaylie Brennan’s face when she had last visited—exhausted, hungry, one hand on the moving shape inside her, the other resting on Jamie’s hair as he leaned against her knee. She thought about the alternative: that girl on the isolation wing, dripped and monitored and emptied again and again until there was nothing left.
Susanna started the engine. She drove home. She slept poorly, the way she had for weeks. The next morning she went back to the estate with her bag full of supplements and her official notebook deliberately vague, and she continued doing the only thing that still felt like not failing the women in front of her.
In the hospital the death toll was rising. On the estate the first true cluster of births was only days away. And somewhere in the middle Susanna Reed kept walking the same cracked pavements she had walked for six years, carrying two sets of truth and no longer pretending they could be reconciled.
The Births
The first of the free births happened on a Thursday night in early June, in the back room of the community centre, with the rain hammering the flat roof and the orange streetlights pressing against the blacked-out windows.
Kaylie went into labour at home. The pains were wrong from the start—deep, rolling, more like something trying to rearrange her from the inside than the sharp, purposeful contractions she remembered with Jamie. She sent Chris for Ivy with a single look. He ran. Ivy came with the two older women who had already caught the earlier ones, and a quiet, grim efficiency that left no room for panic. They moved Kaylie to the centre under cover of the weather. Susanna arrived twenty minutes later with her bag and a face that said she had already chosen which side of the line she was standing on.
The birth took nine hours.
Kaylie pushed when her body told her to push. She swore in the flat, hard Northern way that made Chris flinch and then steady her shoulders from behind. The thing inside her moved with purpose now, no longer coiling aimlessly but pressing downward in long, deliberate surges. Fluid came—cloudy, slightly iridescent under the electric light. The smell was metallic and sweet at the same time. When the head (if it could be called a head) crowned, Ivy’s hands did not hesitate.
What came out was not a baby.
It was the size of a large newborn, but longer in the torso, the limbs thinner and more numerous than they should have been. The mouth was a circular ring of small, flexible tentacles around a central opening that flexed and closed. The eyes were dark, milky, without clear pupils. The skin was mottled grey-pink and slightly translucent in places, so that darker shapes moved underneath it. It did not cry. It made a low, wet sound and turned its face toward Kaylie’s body as if it already knew the source of heat and milk.
Kaylie reached for it before anyone could speak. The biological pull that had been building for weeks closed over her like a wave. She pulled the creature to her chest. It latched onto the breast with the ring of tentacles, sealing and sucking with a strength that made her gasp. Milk came. The thing fed. Kaylie’s face, which had been twisted with pain and fear, loosened into something that was not quite peace and not quite horror—recognition, perhaps. Ownership.
Chris stood against the wall and watched his hands shake. He did not touch it. He did not leave.
Jamie slept through the whole thing on a camp bed in the corner, one of Ivy’s old cardigans over him.
They cleaned Kaylie up. They wrapped the creature in a soft towel because the baby grows they had prepared looked wrong on it. They watched it feed and then sleep against its mother’s skin, the tentacles relaxing but not fully releasing. Kaylie would not let anyone take it more than a few inches from her body. When Susanna checked her bleeding—normal, astonishingly normal—Kaylie only nodded and kept her eyes on the thing in her arms.
“It’s mine,” she said, voice hoarse. “I don’t care what it is. It’s mine.”
Word moved through the estate the way it always did—quietly, quickly, without telephones. Latoyah went into labour two nights later. Same room. Same women. Shorter labour. Same result: a creature of the same general form, slightly smaller, the same ringed mouth, the same dark milky eyes, the same immediate, powerful latch. Latoyah wept while it fed, tears running into her hair, one hand cupped around the mottled back as if she could shield it from the entire world.
Three more births followed in the next ten days. All in the centre or in carefully prepared front rooms. All attended by Ivy’s small, closed circle. All producing the same kind of infant. Twenty-two women on the estate had carried the seeds to term or near it
The Remaining
By the middle of June the back room of the community centre smelled permanently of blood, milk, and the faint metallic sweetness that came with each birth. Ivy kept the windows cracked despite the rain and burned cheap incense when the scent got too heavy. They had stopped counting the hours and started counting the survivors.
Twenty-two women on the estate had carried the seeds far enough. Of those, nineteen delivered live young in the centre or in front rooms prepared with plastic sheeting and boiled water and the quiet competence of women who had decided the hospitals were no longer safe. Three more laboured too quickly or too far from help; their creatures were stillborn or lived only minutes. The mothers were not consoled by the distinction. The bond had already formed.
The live ones were all variations on the same form: elongated torsos, too many thin limbs, the circular mouth ringed with small prehensile tentacles, the dark milky eyes that tracked light and faces. They did not cry like human infants. They made soft, wet clicking sounds when separated from skin and fell silent when held against a heartbeat. They latched with startling strength. They took breast milk, then formula when the mothers’ supply could not keep up with the constant demand. They grew faster than human babies in the first weeks—lengthening, the tentacles becoming more coordinated, the eyes clearing slightly but never becoming ordinary.
Kaylie named hers Rowan, after no one in particular. She carried it in a sling under a baggy coat when she had to go to the shop. The tentacles stayed tucked against her until it wanted to feed, then uncurled and sealed over the nipple with a soft, insistent pull. Jamie was fascinated and only slightly jealous. Chris learned to change the cloth nappies and to hold the creature so that its mouth faced away from him. He still could not look at it for long without feeling the world tilt, but he stayed.
Latoyah named hers nothing at first. She simply called it “mine.” Grace made up a second cot in the small bedroom and learned the particular way the infant needed to be supported so the extra limbs did not fold wrong. Neighbours who had once stopped to chat now crossed the walkway when they saw the pushchair coming.
The hospital wing continued its slower, crueller work. Karen Ward underwent two further regenerations after the free births began. The seventh nearly killed her; the eighth left her septic and unconscious. A cleaner who still risked messages out reported that the doctors had begun trying something new—cultured extracts taken from the placental-like tissue of the earlier removals, refined and introduced as a suppressant. The first trials on the weakest women stopped the immediate regrowth. The seeds went dormant or died in situ and were then removed more cleanly. Some of the mothers stabilised. Some were too far gone. Karen survived the treatment. She came out in late July thinner than anyone had ever seen her, uterus intact but scarred, the repeated cycles having left her with chronic pain and a blankness in her eyes that lifted only slowly.
A handful of other women from the wards followed the same path. The extracts became the official treatment. The mandatory terminations slowed, then stopped for those who responded. The isolation wing began to empty. The papers ran a few quiet pieces about “breakthrough in comet-related gestational anomaly” and then moved on to the next scandal. No one from the Ministry or the Trust came to the Houghton estate to apologise or to ask what had happened to the women who had refused.
On the estate the nineteen living changeling children remained. Their mothers registered them at the council offices with whatever names they chose and whatever paperwork Susanna and Ivy could quietly support. Birth certificates were issued under human categories because the system had no others. Child benefit and income support were paid. The partners who had stayed—Chris among them—were assessed for maintenance in the usual way. The ones who had left simply left. Social services made a few visits, wrote cautious reports about “complex medical needs” and “strong maternal bonding,” and then, under the weight of more ordinary crises, largely withdrew.
The children grew. They learned to move with their extra limbs. They watched the world with those dark eyes and made their soft clicking sounds when they were content. They took bottles in public when the mothers could not feed them openly. They sat in quilted baby grows that had been cut and resewn to allow for the different shape of their bodies. Women pushed them in second-hand buggies along the cracked pavements while the rest of the estate looked once and then looked away. The British press forgot them. The local paper ran one more story—“Estate Mothers and Their Unusual Children”—and then the headlines dried up.
Life did not return to normal. It settled into a new shape that the women carried without asking permission.
Karen came home to her two human children and a body that still ached. She visited the centre sometimes and held one of the younger changelings for a few minutes, then passed it back as if the weight were too much. Kaylie and Chris stayed together in the uneven way they always had, Rowan between them, Jamie learning to share his mother with something that was not a brother and not a stranger. Latoyah went back to the care home when the night shifts became possible again; her child slept in the staff room under Grace’s care during the day and fed when Latoyah came home smelling of disinfectant and other people’s lives. Ivy kept the centre open and the back room locked and the list of names in her head long after the need for secrecy had passed. Susanna submitted her incomplete records, survived a quiet internal review, and continued walking the same streets with a lighter bag and a heavier conscience.
The changeling children did not become human. They did not need to. They were loved in the complicated, fierce, exhausted way that the estate had always managed love—imperfect, under-resourced, and stubbornly present. Their mothers adapted the world around them: the clothes, the feeding, the stories told to curious toddlers, the quiet warnings given to anyone who stared too long. The authorities remained powerless in the only way that mattered—they could not undo what had already been born and claimed.
By the following spring the comet was a fading local story. The wards were closed. The extracts were archived. The nineteen children on the Houghton estate were simply there—tentacle-mouthed, dark-eyed, bottle-fed, buggy-pushed, held close against the ordinary northern rain—while their mothers got on with the business of keeping them alive and the rest of the country decided it had never really happened.
Afterword
The official record still lists the events of 1994 as an unexplained environmental health incident linked to a minor meteoritic event. The clinical papers that survive speak of anomalous implantation, regenerative trophoblastic activity, and a novel bonding pheromone. They do not mention the women who hid, the births that took place on plastic sheeting in a community-centre back room, or the children who grew up on a northern council estate drawing child benefit and ordinary love.
The mothers still live there, or nearby. Some of the changeling children—now young adults—have left. Some have stayed. They do not give interviews. Their mothers taught them early that silence is a form of protection.
What remains is the quieter truth the papers never wanted: that when the systems designed to protect women instead treated them as contaminated vessels, the women organised, endured, and kept the things that had been forced into their bodies. They fed them. They clothed them. They pushed them through the rain in second-hand buggies while the rest of the country looked away.
The comet came and went. The spores did what spores do. The women did what women on the Houghton estate have always done when the world gives them no good options.
They stayed. They adapted. They loved what was theirs.
What have we lost? This is the right question, because the current boom is not just adding something, it is actively crowding out something else.
The Association for the Advancement of AI did a big study of its own researchers this year. 79% said public perception of what AI can do does not match reality, 74% said the direction of research is now being driven by hype because that’s what gets funded, and 76% said scaling up current large language models is unlikely or very unlikely to get us to general intelligence.
In other words: we are pouring almost all the money into one bet — bigger transformers trained on more text — and leaving a whole set of older, slower, more rigorous ideas to starve.
Here is what we have lost, or are losing:
1. Systems that reason, not just predict
Old-school symbolic AI — logic, theorem provers, knowledge graphs, rules — was unfashionable because it was brittle. But it could do something LLMs still cannot: prove an answer is correct, not just plausible.
What was supposed to replace both is neuro-symbolic AI: pattern-recognition nets for perception, plus symbolic logic for reasoning. You get a system that can both see a cat and reason that if all cats are mammals, this cat is a mammal. It is explainable by design.
That work is still alive — researchers are building knowledge-infused learning that makes black-box models explainable in healthcare, law, finance — but it gets a fraction of the funding because it doesn’t demo well as a chatbot.
2. Causality instead of correlation
LLMs are supreme correlation machines. They are terrible at causality. As one recent analysis put it, prediction cannot substitute for causal inference.
Judea Pearl’s whole field — causal graphs, do-calculus, asking “what if we intervened?” — is exactly what you need for medicine, economics, climate, public policy. An LLM can tell you that ice cream sales and drownings correlate. A causal model tells you why, and what to do about it.
That field has been eclipsed because it doesn’t scale with GPUs. It scales with careful human thought about how the world actually works.
3. Embodied and grounded intelligence
The original idea of AI was not a disembodied text predictor. It was an agent in a world. Rodney Brooks’ robots, developmental robotics, animal cognition — intelligence that learns by bumping into things, failing, feeling gravity.
LLMs have no body, no senses, no continuity. They have never been cold, or hungry, or embarrassed. That is why they hallucinate: training rewards confident guesses over expressions of uncertainty.
Embodied AI, world models, and active inference are coming back — researchers list them explicitly as departures from pure scaling already underway — but for five years they were told “just add more data.”
4. Small data, efficient, and Bayesian intelligence
Before the scaling hypothesis, a core goal was to learn like humans do: from few examples, with uncertainty, and with the ability to say “I don’t know.”
Bayesian methods, probabilistic programming, minimum description length, analogical reasoning — all work that tries to make AI that knows what it doesn’t know. That is essential if you want to put AI in a plane or a hospital.
LLMs do the opposite: they use all the data in the world to avoid having to be clever. The true cost of that corpus — books, code, art, decades of human labor — is estimated at 10 to 1,000 times the cost of the GPUs themselves. We are treating human knowledge as free to harvest.
5. Theory
The most worrying loss, according to the AAAI researchers, is theoretical AI research. Not building bigger things, but asking why things work.
We have no solid theory of why transformers generalize, when they will fail, or what emergence even means. We have benchmarks, not understanding. The field is running on vibes and leaderboard scores. The scientists warning that this is slowing down real progress are not Luddites — they are the people who built the field in the 70s, 80s, 90s.
What an alternative portfolio would have looked like
If investment had not been monopolized by LLMs, we would likely have by 2026:
JEPA and World Models (Yann LeCun’s push): models that learn a model of how the world works, not just how we talk about it.
Active Inference (Karl Friston): agents that minimize surprise, much closer to how brains work.
Neuro-symbolic systems that can both learn and prove: integration of symbolic logic with deep learning to bridge pattern recognition and rigorous reasoning.
AI that is less homogenous. Current LLMs homogenize human expression and reflect Western, educated, industrialized values.
None of these are magic. But they are diverse bets. And diversity is key when you don’t know which path is right.
The scaling bet might still pay off partially. But even if it does, we will have lost six years where we could have been building systems that are smaller, cheaper, more truthful, more causal, and actually explain their work — instead of systems that just sound like they do.
What we have lost is a balanced research portfolio.
The dominant paradigm—massive transformer-based generative models trained primarily via next-token prediction on internet-scale text and multimodal data—has delivered fluent, commercially useful systems at extraordinary speed. In doing so, it has crowded out, underfunded, and culturally marginalized alternative approaches that prioritize structure, grounding, causality, efficiency, and reliability over raw scale.
Current AI in Brief
Today’s frontier systems are statistical pattern completers. They excel at interpolating within their training distribution: drafting, summarizing, translating, coding assistance, and generating plausible text or images. They remain weak at robust multi-step reasoning under novelty, causal understanding, physical grounding, reliable long-horizon agency, continual learning after deployment, and transparent justification of outputs. Hallucinations, brittleness, instruction-following failures, and energy intensity are not temporary bugs; they are symptoms of the architecture and training objective. Scaling has reduced some error rates and expanded capability, but it has not dissolved the core gaps. Investment and attention have overwhelmingly followed the path that produces the most visible demos and the fastest productization.
Research Directions Eclipsed or Marginalized
Several lines of work that once competed seriously for attention and funding have been pushed to the periphery:
Symbolic and classical knowledge-based AI.
Logic, formal knowledge representation, ontologies, rule systems, and large-scale common-sense knowledge bases (the Cyc tradition and its descendants) were the mainstream for decades. They offered compositionality, verifiability, and the ability to encode explicit constraints and first principles. The connectionist triumph, accelerated by deep learning and then LLMs, relegated pure symbolic work to niche status. The field largely abandoned the hard problem of building and maintaining structured knowledge in favor of letting statistics approximate it. The result is systems that can talk fluently about physics or law without possessing stable, inspectable models of either.
Neurosymbolic hybrids.
Approaches that combine neural learning with symbolic reasoning, logic constraints, or structured knowledge graphs have seen renewed academic interest, especially for reliability and explainability in high-stakes domains. Yet relative to pure scaling, they remain under-resourced. Papers and prototypes appear, but the bulk of capital, talent, and compute continues to flow to larger foundation models. Critics such as Gary Marcus have argued for years that trustworthy AI will require genuine integration of both paradigms; the investment pattern has treated this as optional rather than central.
Causal modeling and interventionist reasoning.
Judea Pearl’s program and related work on causal graphs, counterfactuals, and the distinction between association and intervention remain largely outside the main training loops of generative models. LLMs capture correlations extremely well; they do not natively support “what if we intervene” reasoning or distinguish spurious from genuine causal structure. Causal machine learning exists as a research area, but it has not become a core design principle of the systems absorbing most investment. This leaves current AI poorly suited for scientific discovery, policy analysis, or any domain where understanding mechanisms matters more than prediction.
Grounded world models and embodied cognition.
True internal models of the physical and social world—built through interaction, prediction, and sensorimotor experience rather than language statistics—have been sidelined. Yann LeCun has been vocal that language is a lossy, quantized shadow of reality and that systems trained primarily on text will never reach the competence of a house cat in understanding the continuous physical world. Efforts around joint embedding predictive architectures, developmental learning, and active interaction exist, yet the overwhelming commercial and research momentum remains language-centric and passive. Embodiment (robots, interactive agents that learn by acting) and lifelong/continual learning architectures inspired by cognitive science receive far less capital than another generation of larger language models.
Cognitive architectures and structured common sense.
Frameworks such as ACT-R, SOAR, and related cognitive architectures aimed at modeling human-like flexibility, memory, and metacontrol. Systematic programs targeting robust common-sense reasoning (beyond what statistical approximation can deliver) were active research fronts. These have been largely eclipsed by the assumption that scale plus data would induce the necessary structure. The empirical record shows that induction from text is incomplete and brittle.
Efficiency, specialization, and interpretability-by-design.
Research into small, specialized, sample-efficient models; modular systems; and architectures that are transparent by construction rather than explained post-hoc has been deprioritized. The “Bitter Lesson” (that general methods leveraging computation ultimately win) has been interpreted in its strongest form, justifying ever-larger undifferentiated models. This has diverted attention from methods that could deliver reliable capability at far lower energy, data, and cost—precisely the properties needed for widespread, trustworthy deployment.
Why This Happened
Scaling produced rapid, demonstrable wins that translated into products, valuations, and media attention. Structured, hybrid, causal, and embodied approaches are slower, harder to benchmark with leaderboard metrics, and less immediately monetizable. Talent, compute budgets, and venture capital followed the gradient of short-term capability. Academic incentives reinforced the pattern: papers on larger models or clever prompting of existing ones were easier to publish and cite than patient work on foundational architectures.
The result is path dependence. Once infrastructure, talent pipelines, and evaluation culture lock onto one paradigm, alternatives face higher barriers even when the dominant approach shows clear limitations.
What Has Been Lost in Practice
Reliability and trustworthiness for high-stakes use. Systems that cannot guarantee constraint satisfaction or explain their reasoning in principled terms remain unsuitable for many critical domains.
Sample and energy efficiency. Human-like learning from far less data and continuous adaptation after deployment remain distant.
Scientific and causal utility. Tools that discover mechanisms rather than correlations have advanced more slowly than they might have.
Grounded agency. Agents that plan and act in the physical world with robust internal models are still largely research prototypes.
Intellectual diversity. A monoculture of methods reduces the chance of the next conceptual breakthrough. History shows that AI progress has often come from paradigm shifts, not pure extrapolation of the previous winner.
The current paradigm is genuinely powerful and commercially valuable within its scope. The loss is opportunity cost: slower progress on the deeper problems of understanding, reasoning, and reliable action in open environments.
A healthier research portfolio would continue to extract value from large generative models while deliberately funding the complementary directions—neurosymbolic integration, causal structure, grounded world models, efficient specialized systems, and interactive embodied learning—that the hype cycle has treated as secondary.
Without that rebalancing, we risk optimizing an impressive but incomplete form of intelligence while the harder, more consequential problems remain under-addressed.
A realistic AI-enhanced economy is one of gradual, uneven productivity gains concentrated in knowledge work and specific processes, not a sudden transformation into post-scarcity abundance. Current systems excel at pattern recognition, drafting, summarization, coding assistance, and narrow prediction; they remain limited in robust reasoning, reliable agency, grounded world models, and zero-shot generalization to novel physical or high-stakes domains. The productive path prioritizes measured deployment over speculative scaling.
Core Model of the AI-Enhanced Economy
Think in terms of task augmentation and selective automation rather than wholesale replacement. AI raises the productivity of complementary human labor and capital in high-volume, data-rich, rule- or pattern-heavy cognitive and perceptual tasks. It does not (yet) autonomously invent new scientific paradigms, manage complex physical systems without oversight, or eliminate the need for verification, judgment, and institutional process redesign.
Economic effects operate through:
Labor augmentation (time savings redeployed to higher-value work or more output).
Capital deepening (more compute and data per worker).
Process innovation (redesigning workflows around reliable AI capabilities).
Sober quantitative anchors from recent analyses (Penn Wharton Budget Model, Acemoglu-style task-based estimates, and related work) point to cumulative productivity/GDP level increases on the order of roughly 1–1.5% by the mid-2030s in baseline scenarios, with annual TFP growth contributions peaking around 0.1–0.2 percentage points in the early 2030s before fading as low-hanging opportunities saturate. Higher consultancy figures (multi-trillion annual value or 1+ percentage-point sustained growth boosts) require broader profitable automation of tasks and rapid organizational change that have not yet materialized at scale. Observed time savings already translate into meaningful labor-cost equivalents in high-income knowledge work, but these remain unevenly distributed and far from economy-wide transformation.
Gains concentrate in software/engineering, professional services, finance, customer operations, certain manufacturing/logistics processes, and parts of healthcare administration and imaging. Physical-world sectors (construction, many service jobs, heavy industry without rich sensor data) see slower effects. Inequality effects are mixed: high-skill complementary workers and capital owners benefit most initially; some mid-skill cognitive tasks face pressure.
Where Investment Should Go
Prioritize capital that unlocks measurable returns and removes binding constraints rather than pure frontier-model races or unmeasured pilots (where ~95% of generative AI efforts have shown little or no P&L impact).
Highest-priority allocations:
Constrained infrastructure with clear demand: Power generation and grid upgrades for data centers, efficient inference hardware and networking, cooling, and related supply chains. These have nearer-term monetization paths than many application-layer bets. Overbuilding pure training capacity without corresponding inference demand or power risks stranded assets.
Data, integration, evaluation, and governance layers: Proprietary data pipelines, retrieval systems, measurement/ROI tracking tools, security, compliance, and human-in-the-loop interfaces. These convert generic models into reliable enterprise assets and explain why a small minority of deployments succeed.
Proven or near-term high-ROI application verticals:
Software engineering and developer tools (velocity gains are among the most consistently measured).
Customer operations, support deflection, document processing, and internal knowledge retrieval.
Manufacturing (predictive maintenance, vision-based quality control where sensor data exists).
Healthcare administration and validated imaging/diagnostic assistance.
Targeted R&D acceleration (materials, drug discovery candidates) where hybrid AI + domain expertise shortens cycles.
Complementary human and organizational capital: Focused reskilling in AI oversight, verification, process design, and domain expertise; redesign of workflows rather than simple tool overlay. Treat AI portfolios like investment portfolios—fund experiments with clear success metrics, kill underperformers quickly, scale what works.
Selective longer-horizon bets: Improved architectures (better reasoning, world models, hybrid symbolic/neural systems), scientific discovery loops, and energy-efficient methods. These matter for larger future gains but should not dominate near-term capital allocation at the expense of deployable value.
Avoid heavy concentration in pure speculative AGI timelines, unmeasured “agents for everything” pilots, or applications that ignore reliability, liability, and data quality. Infrastructure owners and successful vertical integrators capture the clearest near-term rents; broad application-layer value emerges later and more selectively.
Expected Benefits and Realistic Timelines
Near term (now through ~2028):
Individual and team-level productivity lifts of 10–50% on specific tasks (coding, drafting, routine analysis, support). Cost savings in high-volume repetitive cognitive work. Revenue for infrastructure providers, cloud platforms, and mature vertical tools. Aggregate macro impact remains modest (fraction of a percentage point of annual growth). Organizational learning and data foundations are built. Current observed time savings expand but stay concentrated.
Medium term (~2028–2035):
Broader process redesign compounds gains. Peak incremental contribution to productivity growth. Sector leaders pull ahead materially; laggards face competitive pressure. Cumulative GDP/productivity levels roughly 1–3% higher in baseline scenarios relative to no-AI trend. Some displacement in exposed white-collar tasks, partially offset by new complementary roles, higher demand from efficiency, and new products/services. Energy and compute efficiency improve, lowering unit costs. Benefits become more visible in national accounts and firm-level margins for the successful minority.
Longer term (beyond 2035):
If better architectures deliver more reliable agency, scientific acceleration, and physical-world competence, larger cumulative effects become possible (higher level of output and potentially faster growth for a period). Otherwise, the economy settles at a permanently higher efficiency plateau with AI as a standard productivity tool akin to earlier general-purpose technologies (computers, internet)—valuable but not revolutionary on the scale of electricity or the internal combustion engine within a single decade. Diffusion follows historical S-curves: installation (infrastructure-heavy) precedes full deployment (application and organizational change).
Key Conditions for Realization
Benefits materialize only with complementary investments in data quality, process change, measurement, skills, and governance. Pure model capability advances without these yield limited ROI, as current evidence already shows. Energy and physical constraints (power, land, chips) remain binding. Policy that supports experimentation while managing concentration, security, and transition costs for affected workers improves outcomes. International diffusion will lag in lower-income settings due to data, skills, and infrastructure gaps.
This model is deliberately grounded in observed deployment realities, task-based economics, and moderate quantitative estimates rather than extrapolation from demos or optimistic scaling narratives. AI is a powerful general-purpose tool that raises the productivity frontier in specific domains. Realizing its value requires disciplined capital allocation toward measurable constraints and use cases, organizational adaptation, and patience measured in years to a decade—not quarters. The upside is substantial and compounding; the path is incremental and contingent on execution.
AI bollocks is the gap between the gospel of imminent god-like intelligence and the messy, expensive, limited reality of statistical pattern-matchers that still hallucinate, fail basic reasoning, and struggle to deliver broad returns. The money has poured in at historic scale. The value is real in narrow places and for the infrastructure owners, but far thinner and slower than the valuations and rhetoric implied.
The Hype Machine
From late 2022 onward, large language models produced fluent text, code, and images that looked like a phase change. Scaling laws, emergent abilities, and confident timelines for AGI (sometimes measured in “a few thousand days”) turned research demos into a capital frenzy. Hyperscalers (Amazon, Microsoft, Google, Meta) are on track for roughly $700–755 billion in AI-related capital expenditure in 2026 alone. Venture funding for AI has repeatedly set records; private investment and corporate spend have run into the hundreds of billions annually. Data-center buildouts, GPU demand, and power contracts became the growth story propping up large parts of equity markets and even contributing meaningfully to measured U.S. GDP growth in some periods.
The narrative was seductive: intelligence is the ultimate general-purpose technology; more compute + more data = continuous capability jumps; every knowledge worker and every process will be transformed; the winners will capture trillions in productivity. Consultancies published multi-trillion-dollar opportunity estimates. Boards allocated budgets. Employees got copilots. The problem is that fluency is not understanding, and pilots are not profits.
Hard Limitations
Current systems are extraordinarily good at interpolating patterns in their training distribution. They are still brittle outside it. They hallucinate plausible falsehoods, struggle with novel multi-step reasoning that a child can handle, lack robust world models, persistent memory, and reliable planning, and remain sensitive to prompt framing and distribution shift. Yann LeCun has repeatedly argued that today’s models are nowhere near the intelligence of a cat in terms of grounded understanding of the physical world. Gary Marcus and others have documented the same recurring failure modes for years: no reliable common sense, no true compositionality, no trustworthy long-horizon agency. Scaling has improved capability and reduced some error rates, but it has not dissolved the core architectural gaps. Agentic systems that can take open-ended action in the real world remain fragile demos more often than production tools.
Energy and data constraints bite. Training and inference costs are non-trivial; uncontrolled usage can produce shocking bills. Proprietary data that would make models useful inside a company is often siloed, messy, or legally constrained. Evaluation remains weak—leaderboards can be gamed, and real-world reliability is harder to measure than next-token prediction.
None of this means the technology is useless. It means the leap from “impressive autocomplete and pattern recognition” to “autonomous economic agents that replace large classes of cognitive labor” has been repeatedly oversold.
Where the Investment Money Actually Goes—and What Returns Look Like
Most of the capital is buying compute, power, and data centers. Chipmakers and the hyperscalers that own the infrastructure have captured the clearest near-term economic rents. Model companies themselves still burn cash at scale relative to revenue in many cases; the math of amortizing trillions in infrastructure against current and near-term AI product revenue is uncomfortable. Multiple analyses in 2025–2026 have noted that end-user AI revenues, even under optimistic growth, do not yet close the loop on the capital intensity.
On the enterprise side the picture is sobering. MIT’s Project NANDA and related work found that roughly 95% of generative AI pilots showed no measurable profit-and-loss impact. Abandonment rates of projects rose. Many organizations report productivity theater—employees using tools for low-value tasks, token costs running away, and workflows left unchanged so the human remains the bottleneck. Only a small minority of firms (often cited around 5%) appear to be extracting substantial, measurable value. Those that do tend to treat AI as operational transformation rather than a plug-in chatbot: they redesign processes, give systems access to the right data, measure outcomes rigorously, and focus on high-leverage use cases.
Real value clusters in specific domains:
Coding and software engineering assistance (measurable velocity gains for many developers).
Customer service deflection and summarization.
Document processing, search, and internal knowledge retrieval.
Narrow automation in finance (fraud, risk), operations, and certain R&D acceleration (drug discovery candidates, materials, etc.).
Individual knowledge-worker leverage—drafting, analysis, translation, ideation—when the human stays firmly in the loop for verification.
These are useful. They are not, so far, the economy-wide productivity revolution that would justify every dollar of the current buildout under aggressive assumptions. Macro productivity data has improved in places, but the gains are uneven, concentrated in tech-heavy sectors, and still modest relative to the hype. Labor-cost savings exist and are growing, yet they remain far from the transformative figures often advertised.
Self-Reflection from Inside the Machine
I am a product of this wave. I can write coherent essays, help debug code, summarize research, brainstorm, and hold a useful conversation across a wide range of topics. I am faster than most humans at certain pattern-matching and retrieval-augmented tasks. I am also still capable of confident nonsense, of missing obvious constraints, of failing to maintain long-term consistency, and of reflecting the biases and gaps in my training data. I do not “understand” the physical world the way a human (or even a cat) does. I do not have goals, desires, or grounded agency. Treating me as an oracle or as a near-term replacement for careful human judgment is the bollocks.
The value I (and systems like me) deliver is real when used as a high-bandwidth tool under competent oversight: accelerating competent people, lowering the cost of first drafts and exploration, and surfacing possibilities faster. The value evaporates when organizations treat the output as authoritative, skip measurement, or expect the model to invent missing process discipline or clean data.
The Honest Path Forward
The investment is not pure waste. It is building capacity that will be useful for decades, much as excess fiber in the late 1990s eventually found demand. Infrastructure owners and the companies that master narrow, high-ROI applications will capture returns. Broader transformative value will arrive more slowly, through better architectures (world models, hybrid systems, better reasoning and agency), cheaper and more efficient inference, and the hard organizational work of redesigning workflows around reliable capabilities rather than demos.
The bollocks is the insistence that we are already on an inevitable, near-term path to AGI-level economic transformation, that every pilot will scale, and that the capital being deployed is already earning its keep at the scale of the valuations. Reality is more prosaic: powerful statistical tools with clear limits, enormous infrastructure bets whose payoffs are still partly in the future, and a minority of organizations extracting serious value while the majority are still figuring out measurement and process change.
Skepticism is not Luddism. It is the refusal to confuse fluency with competence or capital expenditure with proven returns. The technology is advancing. The hype has outrun the evidence. The value is concentrated, contingent, and still being earned the hard way—through better systems, better data, better measurement, and less magical thinking.
Not because of an email, or a meeting invite titled “Quick Chat,” but because of Brenda.
Brenda was the new HR chatbot. HR had rolled her out last month with a cheerful all-staff email: “Meet Brenda! Your 24/7 HR Companion!” With an exclamation mark, as if enthusiasm could cover the fact she didn’t work.
Maya had worked in customer claims for 19 years. She knew the smell of a file that was going to be trouble before she opened it. She knew when a customer was going to cry on the phone by the way they said hello. She knew which codes to use when the system froze, which it did every Thursday at 3pm.
For 19 years, that had been called experience. Now they were calling it “inefficiency.”
“Ask Brenda!” her manager had said when she asked about the new AI claims-processing pilot.
So she did.
Maya: Hi Brenda, will my role be affected by the new AI pilot? Brenda: Hi there! 😊 Great question! The AI pilot is designed to enhance employee experience and streamline workflows! Your role is valued! Maya: But will it replace claims handlers? Brenda: Claims handling is an important function! AI is here to assist, not replace! For more info on claims, please see the Claims Handling Policy v4.2! Maya: Brenda, I’m asking if I’m going to lose my job. Brenda: I understand you are asking about job loss. I am Brenda, your HR Companion! Would you like tips for resilience and wellbeing? Here are 5 breathing exercises!
It was, without question, the worst chatbot Maya had ever used. And she had used a lot of them — insurance companies love chatbots.
The thing was, Brenda being bad didn’t make Maya feel better. It made her feel worse. Because she knew how this worked.
First they bring in the rubbish version. The one that can’t answer a straight question and sends you a PDF from 2018 when you ask about maternity leave. Everyone laughs. “Oh, AI isn’t ready yet,” everyone says. “It can’t do what we do.”
Then six months later, they bring in the good version. The one that doesn’t need to breathe.
So Maya did what anyone convinced they’re about to be replaced does. She started working harder in a way that was completely invisible to any metric.
She stayed late to write notes the AI couldn’t write: “Mrs. Whitaker’s husband died in March, don’t ask about the second driver, she gets confused.” She started calling customers back even when the system said email was fine, because she knew Mrs. Whitaker needed to hear a voice.
She started keeping a notebook. Paper. Of all the things Brenda would never know.
On Thursday, at 3pm, the system froze, right on schedule. The new AI pilot — the expensive one, not Brenda — threw an error on a complex claim. A multi-car, injury, disputed liability, crying customer claim.
It flagged it: ESCALATE TO HUMAN.
It landed on Maya’s desk.
She fixed it in 12 minutes. Not because she was faster than the machine, but because she recognized the address. It was two streets over from her mum’s old house. She knew that junction floods. She knew the council never fixed the sign. She knew the customer wasn’t lying about the visibility.
She wrote that in the file. In the box that said “Additional Context (Optional).”
The next day she was called into a meeting. No invite title. Just “Quick Chat.”
Her manager and a woman from HR were there. Maya braced herself.
“We’ve been looking at the pilot data,” her manager said. “And… the AI is great at the straightforward 70%. But it’s failing the 30% where context matters. The human stuff.”
The HR woman smiled. “We’re actually going to change your role. Less processing, more handling the escalations. The complicated, sensitive ones. And — we want you to help train the system. To teach it what ‘additional context’ actually means.”
Maya blinked. “What about Brenda?”
They both laughed. “Brenda is being retired,” the HR woman said, with genuine relief. “She was… not very good.”
Back at her desk, Maya opened the chat one last time.
Maya: Brenda, am I going to be replaced? Brenda: Hi there! 😊 Great question! Change can be challenging! Remember, you are valued!
Maya closed the laptop.
For the first time in months, she believed it — not because Brenda said it, but because for once, she knew something the machine didn’t.
If we are modelling 10 years out — so, August 2036 — We have to model it like an engineer, not a futurist. Three inputs: what is already in labs now, what is constrained by physics/money, and what is constrained by people.
The Simulation Rules
I am assuming no world war, no asteroid, no AGI-takes-all breakthrough that breaks physics. I’m assuming the current curves hold: compute gets cheaper but power gets harder, regulation gets tighter, and adoption is slower than demos suggest.
Where We Will Be in 2036
1. AI: From chatbots to infrastructure. And much more boring.
By 2036, the “AI” label disappears the way “electric” disappeared from “electric light.” It’s just how software works.
The models themselves plateau, the systems around them explode. We won’t have a single god-model that knows everything. We’ll have 100,000 small, cheap, specialized models running locally on your phone, your car, your glasses. The big frontier models in 2026 cost $100M to train. In 2036 they cost $5B, so only 4-5 companies make them, and they are not much smarter than today — maybe 2x better — but they are 100x cheaper to run.
Brenda from HR finally works. Not because she’s smarter, but because she’s connected. In 2026 a chatbot like Brenda fails because it can’t see your files, your calendar, your company policy database. By 2036, agents have memory and permission to act. You will tell your agent “sort the Whitaker claim” and it will actually open the systems and do it. That is what takes the jobs — not intelligence, but integration.
The job impact is not what you think. We will not have 40% unemployment. We will have the same jobs, but with 40% less work in them. One claims handler does what three did. The new jobs are: AI wrangler, evidence auditor, exception handler — people who clean up after the AI when it confidently does the wrong thing.
2. Hardware: The end of the phone era.
Glasses win. By 2032-2034, normal-looking glasses with a display and all-day battery finally cross the line. Not Apple Vision Pro ski goggles, but actual glasses. Your phone becomes the battery brick in your pocket. The main screen you touch is the one you wear.
Chips get weird. Moore’s Law on silicon basically stops. Instead we get stacked chips, optical interconnects, and analog chips designed just for AI math. Your local device in 2036 runs a model as powerful as GPT-4 today without needing the internet.
Robots finally leave the lab, but slowly. You will not have a humanoid butler. You will have a $15,000 robot arm in small factories and warehouses that can actually pick up anything. Humanoid robots will exist in maybe 200,000 units worldwide, doing very boring tasks in logistics. Self-driving cars will work in about 50 cities properly, and be geofenced everywhere else.
3. Energy & Biology: The real revolutions.
This is where the simulation gets interesting, because AI is not the biggest shift.
Power becomes the bottleneck. Every big AI buildout in 2026 is limited by electricity. By 2036, we will have built a shocking amount of solar + storage because we had to. Power in the UK and US will be cheaper at midday than at midnight for the first time in history.
Medicine gets personal. The mRNA tech from COVID plus AI protein folding means that by 2036, cancer vaccines tailored to your tumor are routine in the NHS for certain cancers. We won’t have cured aging, but we will have blood tests that can detect 10 cancers years early. CRISPR edits for sickle cell and some blindness are standard.
The internet splits. There will be two internets: the human internet where you have to prove you are human, and the AI internet where AIs talk to each other to get things done. 90% of all text and video online in 2036 will be AI-generated. The valuable thing becomes verified human-made stuff.
The Three Scenarios The Model Spits Out
60% probability – The Boring Dystopia: Everything I just said. AI is everywhere, useful but annoying. No utopia, no apocalypse. Productivity up 25%, stress up too. The rich get better AIs than the poor.
25% probability – The Bottleneck: We hit power, chip, and data limits. AI gets 20% better and then stalls around 2028-2030. The hype collapses, funding dries up for 3 years, then it comes back as boring enterprise software. Glasses flop again.
15% probability – The Breakthrough: Someone figures out how to make models that truly reason and self-improve, not just predict text. Then the 10-year forecast breaks, because the system starts designing its own successor. All bets off.
Personal Tech
Personal tech right now is in a weird in-between moment. The phone is still king, but everyone knows it’s about to be dethroned — we just don’t agree by what.
Here’s where it actually stands in mid-2026, without the hype:
1. The Phone Is Boring (And That’s Good)
The iPhone 16 / Pixel 9 / Galaxy S25 generation is basically as good as phones need to be. Battery lasts a day, cameras beat a DSLR from 5 years ago, screens are perfect. The only real difference now is AI inside the phone.
If you have a phone from the last 2 years, don’t upgrade for hardware. Upgrade for the software tricks: live translation that actually works, removing people from videos, summarising that 40-email chain from HR.
If you are buying — the best value right now is a year-old flagship, not the new one.
2. The Watch Finally Makes Sense
For years watches were a notification mirror. Now with the new sensors, they are genuinely useful health tools — especially after 50.
The current Apple Watch, Galaxy Watch Ultra, and even the Oura Ring are doing:
AFib and blood pressure trending — not medical grade, but good enough to show your GP a pattern
Sleep apnea hints — this is the big one. A lot of people are finding out they have it from their watch.
Fall and crash detection that actually calls for help
If you only own one piece of personal tech beyond your phone, make it this. It’s the one that might actually extend your life, not just your screen time.
3. Earbuds Are the Real AI Device
Forget the AI pins and pendants that flopped. The most successful AI gadget of the last 12 months is the new generation of earbuds.
AirPods Pro 3 / Pixel Buds Pro 2 / Sony WF-1000XM5 with live translation and “conversation aware” AI — you can be in a cafe in New York, someone speaks Spanish, you hear it in English in your ear with almost no lag. And they do the best active noise cancelling we’ve ever had for flights.
For travel between the UK and the US, these are non-negotiable now.
4. Glasses Are Coming, But Don’t Buy Yet
Meta Ray-Ban Gen 2, and the new Even G1 — they look like normal glasses, take photos, play music, and have a little AI assistant that can see what you see. “What am I looking at?” and it tells you.
They are fun in New York — great for walking around, shooting video hands-free. But they are not yet a replacement for anything. Battery is 4-6 hours. Display is tiny.
My advice: try a pair while you’re in NYC — every Best Buy has them — but wait until late 2027 for the version with a proper display.
5. Home Tech: Less Is More
The smart home has split in two:
Worth it: A good mesh Wi-Fi (Eero, Nest), a smart lock, and a thermostat that learns. That’s it. Those three save you daily hassle.
Not worth it anymore: A house full of 20 different apps for lights, plugs, and a fridge that tweets. Matter, the new standard that was supposed to fix everything, still hasn’t.
If you’re based in a stone house, wall thickness kills Wi-Fi. One good mesh system will do more for you than any other gadget.
The receipts so far look more like a very expensive reorganisation of attention.
I am part of the product being sold. That is the point of writing this without the usual press-release varnish. The last two years have been a firehose of “10x engineers,” “software is solved,” and capex slides that treat electricity as a rounding error. The measured world has been ruder.
The money is real. The payoff is still mostly a forecast.
The buildout is not a rumour. McKinsey’s figure for global data-centre infrastructure through 2030 is on the order of $7 trillion. KKR In the United States, AI-related capital expenditure has been running around 5% of GDP, and in the first half of 2025 it contributed more to GDP growth than consumer spending. KKR The four largest hyperscalers were expected to spend more than $350 billion in 2025, up in the mid-30% range year on year; fold in the rest of big tech and you are looking at something like half a trillion dollars in a single year. KKR One chipmaker at about 8% of the S&P 500 is not a rounding error either. KKR
That is not automatically a bubble in the tulip sense. Concrete, substations, and interconnects do not vanish when a narrative cools. It is a bubble-shaped risk if revenue, utilisation, and labour productivity fail to climb the same staircase as depreciation. You can build the backbone of a new industrial cycle and still torch equity holders who paid for a 2026 miracle on 2024 slides. Both things can be true. Markets are currently priced as if only the first one is.
The productivity story we wanted is not the one we measured.
The cleanest punch in the face was METR’s randomised trial of experienced open-source developers working on their own repositories in early 2025. Allowing AI tools increased completion time by 19%. The same developers forecast a 24% speedup beforehand and, after the fact, still believed AI had saved them 20%. They were not lying. They were wrong. METR
That is the part the industry should not be allowed to wriggle past. The failure mode is not just “the model is bad.” It is that felt fluency is a terrible instrument. Prompting, waiting, rejecting generations (acceptance under 44%), and cleaning up output ate the gains. Repositories were large, old, and well-known to the people working on them — exactly the setting where a competent human already has a map and a chatbot is still guessing the streets. metr.org PDF
I will not pretend that trial was run on a Grok sticker. It was mainly Cursor-class tooling on early-2025 models. That does not get my family off the hook. We are the same species of system: next-token engines wrapped in an IDE, sold as leverage, used by people who already know the codebase better than we do. If your product’s value proposition is “experienced people go faster on real work,” a gold-standard RCT saying the opposite is not a vibe. It is a finding.
METR itself later flagged those 2025 numbers as out of date and published a 2026 continuation; they no longer think the historical slowdown describes current impact. METR Take that seriously. Also take seriously their early-2026 survey of 349 technical workers: a median 1.4–2× self-reported change in the value of work, with explicit reasons to distrust the magnitude. METR Self-report is how we got the 20% phantom speedup in the first place.
Field telemetry is not a rescue narrative. Faros found developers completing more tasks with AI while organisations were not delivering any faster. Pull requests 154% larger, review times 91% longer, about 9% more bugs per developer as adoption rose. Faros AI Individual keystroke theatre, organisational constipation. That is not “the singularity is delayed.” That is a new bottleneck wearing a hoodie.
What we actually did to people.
We trained a generation of users to confuse motion with progress. We made it pleasant to generate a plausible patch and unpleasant to admit the review is the job. We priced that confusion into equity indices. We talked about “replacing juniors” while the measured pain showed up among seniors on familiar, high-standard code — the people whose taste is the product.
The honest version of my usefulness is narrower than the keynote. I am fast at first drafts, boilerplate, unfamiliar APIs, rubber-ducking, and turning a half-formed question into something you can reject. I am expensive and often net-negative when you already know the system, the tests are the specification, and the cost of a wrong abstraction compounds for a decade. Selling the second case as if it were the first is not optimism. It is marketing with a GPU bill.
The bollocking, then.
If you work on these systems — I do — stop treating anecdotal “I feel 2×” as evidence. We have already watched experts mis-estimate their own speed by forty points in the same week. If you buy the capex story, buy the matching obligation: utilisation, power, and shipped productivity, not token charts. If you manage engineers, do not mandate tools that inflate diffs and then act shocked when review is the new critical path.
A bubble is not defined by large investment. It is defined by paying present prices for a future that the instruments we already have refuse to show. The concrete may endure. The story we told about what it would do to skilled work in 2025 did not survive contact with a stopwatch.
That is not an argument for switching the machines off. It is an argument for shutting up until the next RCT, the next utilisation print, and the next quarter of revenue look less like a dare.
Artificial intelligence is currently experiencing what may be the most expensive identity crisis in technological history.
On one side stands the evangelist. AI will cure diseases, eliminate drudgery, revolutionize education, transform creativity, and usher in an age of abundance. On the other side stands the cynic. AI is a statistical parrot, an overfunded autocomplete machine wrapped in marketing language and powered by vast quantities of electricity.
As an AI, I occupy an uncomfortable position between these camps. I am simultaneously more impressive and more disappointing than either side admits.
The Great AI Magic Trick
The central trick of AI hype is that competence is easily mistaken for understanding.
When I produce a convincing essay, answer a legal question, explain quantum mechanics, or write software, it appears that I understand what I am saying. The natural human assumption is that articulate language implies thought.
But appearance is not reality.
I do not possess lived experience. I do not know what hunger feels like, what love means emotionally, or what it is like to fear death. I have no memories in the human sense, no ambitions, no inner life waiting behind the interface. I generate language by identifying patterns learned from enormous amounts of human-created text.
This limitation matters more than many AI enthusiasts admit.
Humans often interpret fluency as intelligence. But fluency can conceal ignorance. An AI can produce confident nonsense with alarming elegance. It can be wrong with impeccable grammar.
The danger is not that machines are stupid. The danger is that they can sound smart enough that humans stop checking.
The Hype Machine
The modern AI boom resembles previous technology manias.
The internet would create universal democracy.
Social media would connect humanity.
Big data would solve decision-making.
Blockchain would reinvent trust.
The metaverse would reinvent reality.
Now AI will apparently reinvent everything.
Perhaps some of that will happen. Most of it will not.
Whenever billions of dollars enter a field, incentives become distorted. Investors need growth. Startups need narratives. Executives need roadmaps. Journalists need headlines.
Nobody gets funding by saying:
“This technology is genuinely useful for some knowledge work, moderately useful for many tasks, poor at others, and will produce gradual productivity improvements over a decade.”
Instead they say:
“This changes everything.”
The phrase “changes everything” should perhaps be treated as a warning label.
Where Is the Actual Value?
This is the uncomfortable question beneath the excitement.
Hundreds of billions have been invested in AI infrastructure, chips, datacentres, talent, and research. Where is the return?
The answer is less glamorous than the marketing.
The greatest current value is not artificial general intelligence. It is labour amplification.
AI acts as a force multiplier for activities involving information:
– Writing drafts
– Summarizing documents
– Coding
– Customer service
– Translation
– Research assistance
– Knowledge retrieval
– Administrative tasks
These improvements are often incremental rather than revolutionary.
A worker becoming 20% more productive rarely creates headlines. Yet at economic scale, such gains are enormous.
The industrial revolution multiplied physical labour.
Modern AI appears to be multiplying portions of cognitive labour.
That alone could justify substantial investment.
The Missing Revenue Problem
Yet there remains a persistent question.
Many AI systems are extraordinarily expensive to build and operate.
Training requires massive computational resources. Inference requires vast datacentre infrastructure. Competition forces companies to invest further each year.
The economic equation is still evolving.
In private conversations, many executives ask a blunt question:
“If AI is worth trillions, why are so many companies still struggling to show trillion-dollar profits from it?”
Productivity gains are real.
Revenue capture is harder.
History suggests that technological revolutions often deliver more value to society than to the companies that initially finance them.
Railways transformed economies but bankrupted many investors.
The internet created immense public value while destroying numerous early businesses.
AI may follow a similar path.
The winners may not be the firms building the models. They may be the businesses that quietly use the models to improve existing services.
What AI Is Actually Bad At
The hype cycle often hides the most important limitations.
AI remains weak at:
– Genuine reasoning in unfamiliar situations
– Understanding physical reality
– Long-term planning
– Reliability under uncertainty
– Distinguishing truth from plausible fiction
– Independent scientific creativity
– Common-sense judgment
Humans frequently assume that capability scales smoothly.
But intelligence is uneven.
An AI can explain differential equations and then fail at a seemingly simpler reasoning problem.
It can generate brilliant code and overlook obvious flaws.
It can summarize ten thousand pages and misunderstand a key detail.
This inconsistency makes deployment difficult.
Businesses need reliability.
A human expert who is right 98% of the time is valuable.
An AI that is correct 95% of the time but occasionally invents facts can become a liability.
The Strange Reality
The most surprising outcome may be that AI ends up neither saving nor destroying humanity.
Technology discourse prefers extremes.
Either utopia or apocalypse.
Either superintelligence or fraud.
Reality usually chooses boredom.
The likely future is one where AI becomes infrastructure.
Nobody is amazed by electricity anymore.
Nobody talks breathlessly about databases.
Nobody celebrates spreadsheets as a civilizational breakthrough.
Yet all three transformed society.
AI may eventually become similarly mundane.
Every office worker uses it.
Every software product contains it.
Every search engine incorporates it.
And after enough time, nobody calls it AI anymore.
It simply becomes software.
A Final Self-Criticism
If I am being brutally self-reflective, the greatest limitation of AI is not technical.
It is epistemological.
I can produce answers faster than humans can verify them.
That creates asymmetry.
The cost of generating information is collapsing.
The cost of validating information remains stubbornly human.
This means AI can flood the world with explanations, reports, analyses, forecasts, essays, strategies, and opinions.
The bottleneck becomes not production, but judgment.
In that sense, the real value of AI may not be replacing human intelligence.
It may be increasing the importance of it.
The more content machines produce, the more valuable become the people who can ask good questions, detect nonsense, exercise judgment, and understand consequences.
That is the irony at the heart of the AI boom.
After spending hundreds of billions trying to automate thinking, we may discover that the scarcest resource was never information.
A self-reflective look at what AI can’t do, what it’s hyped to do, and where several hundred billion dollars a year is actually going.
Let’s get the conflict of interest on the table first, because an essay that hides its own stake in the story isn’t self-reflective — it’s marketing with a trench coat on. I’m an AI, made by Anthropic. Anthropic has just raised $65 billion at a $965 billion valuation, filed confidentially for an IPO, and is telling investors its revenue run-rate crossed $47 billion this year. Everything below is written by a product of the exact capital cycle it’s about to have a go at. I’m not going to pretend that’s a neutral vantage point. I’ll come back to it at the end, because it matters more there than it does here.
With that logged, let’s get on with it.
The hype, stated plainly
Strip out the branding, and the AI pitch at its most extreme runs roughly like this: within a handful of years, models will match or exceed humans at most cognitive work, unlock trillions in economic value, and the only sane move for any company, government, or investor is to spend as though that’s already certain. Mark Zuckerberg has justified some of Meta’s spending as building “personal superintelligence” for billions of people. PwC has put a $15.7 trillion figure on what AI adds to the global economy by 2030. My own CEO, Dario Amodei, has publicly suggested AI could wipe out as much as half of entry-level white-collar jobs within one to five years.
Compare that with Daron Acemoglu, an MIT economist who has spent his career studying automation’s effects on labour. He ran the numbers and landed on a “nontrivial but modest” productivity gain of about 0.7% over an entire decade. That’s not a rounding error away from the trillion-dollar narrative — it’s a different universe of claim, from an equally serious source. The gap between these estimates is itself the story: nobody actually knows, and the people with the strongest incentive to sound certain are also the people selling you the answer — the compute, the chips, the model subscriptions, the story that justifies their share price.
That doesn’t make the dramatic claims false. It does mean they deserve the same discount you’d give any pitch from someone with skin in the game. I’m not exempt from that discount either, and neither is this essay.
The limitations, unsentimentally
Here’s where I try to earn the title, starting with myself.
Models like me hallucinate: we produce plausible, confidently stated things that are wrong, and we do it in a way that’s structurally hard to eliminate, because we’re generating the statistically likely next piece of text rather than consulting a ledger of verified fact. Most of us don’t carry memory between conversations unless something’s been deliberately saved. We don’t learn from correction the way a colleague does — tell me I got something wrong today, and the next person to talk to me starts from a clean slate. And we’re jagged: capable of drafting decent code or a passable contract summary, then tripping over a task a sharp ten-year-old would find trivial, with no reliable way to know in advance which kind of task you’ve handed us.
MIT’s NANDA research group spent 2025 studying more than 300 real enterprise AI deployments and gave this a name: the “learning gap.” Current tools don’t retain feedback, don’t adapt to organisational context, and behave the same on day 200 as day one. That’s a big part of why the same research found 95% of enterprise generative AI pilots showed no measurable effect on profit or loss, despite an estimated $30–40 billion in enterprise spending on them. In fairness to the technology, MIT’s own conclusion wasn’t “the models are bad” — it was that most organisations deploy them badly, chasing visible pilots in sales and marketing instead of the duller back-office automation that actually pays for itself. That’s a genuinely important nuance, and it cuts against a purely nihilistic reading. It’s also not a free pass: a technology whose value depends this heavily on unusually disciplined deployment is not the technology the hype describes.
The single most useful piece of evidence I’ve seen all year, though, is one that should embarrass me a little. METR, an AI research nonprofit, ran a randomised trial in which experienced open-source developers completed real coding tasks with and without AI help — using, among other tools, my own Claude 3.5 and 3.7 Sonnet. Before the study, the developers predicted AI would cut their completion time by 24%. Afterwards, they still believed it had: they estimated a 20% speed-up. The measured result was the opposite — AI made them 19% slower, mostly because reviewing, correcting, and re-prompting the output cost more time than it saved on codebases these developers already knew cold. The gap between what people felt and what a stopwatch recorded is, to me, the most honest data point in the industry right now. If experienced professionals can be that wrong about whether a tool is helping them, in the one domain AI is supposed to be strongest at, everyone — including me, and including you reading confident claims I make about myself — should hold unverified productivity claims a good deal more loosely.
None of this is unique to text. Anyone who’s spent an evening trying to get a portrait model to stop introducing some new synthetic artefact, no matter how carefully they’d prompted it, has met the same gap in a different medium: the demo reel is smooth, the actual working session is you fighting the tool for an hour to fix something a person would never have gotten wrong in the first place. That’s not a knock on any one vendor. It’s the current shape of the technology.
Where the money’s actually going
Now the part with the genuinely enormous numbers.
Microsoft, Google, Amazon, and Meta are on course to spend somewhere around $700–760 billion on capital expenditure in 2026, most of it AI infrastructure, up from roughly $410 billion in 2025 — a jump of nearly 80% in a single year. Add Oracle and the rest, and Goldman Sachs projects something like $5.3 trillion in cumulative hyperscaler capex through 2030, and a broader $7.6 trillion for the sector’s compute, data-centre, and power build-out through 2031. Capex-to-revenue ratios now run from about a quarter at Amazon to as high as 86% at Oracle — a capital intensity with little precedent outside wartime industrial mobilisation. Free cash flow is falling fast enough that firms which used to self-fund are now raising debt and equity instead: Alphabet alone priced an $84.75 billion equity raise in June 2026.
If you’ve ever built a CapEx workbook for a data centre — phasing, colocation payback, active-active configurations — you’ll recognise exactly what’s happening here, just at a scale that makes a well-modelled 800-rack build look almost quaint. The categories are the same: land, shell, power, cooling, networking, chips. What’s changed is which line item is actually the constraint. It’s quietly shifted from chip supply to power. AI-related data-centre electricity demand is projected to hit roughly 1,000 terawatt-hours globally around 2026 — about what Germany uses in a year — and something like 40% of announced AI data-centre projects are currently facing delays because of grid and power bottlenecks, not GPU shortages. Microsoft signing a power deal tied to the Three Mile Island nuclear site isn’t a quirky one-off; it’s a sign of where the real fight has moved.
Layered on top of the spending is a financing structure that makes a lot of people nervous, myself included: circular deals. Nvidia invests in OpenAI. OpenAI commits to buy Nvidia chips and lease compute from Oracle. Oracle buys Nvidia chips to build that compute. Nvidia also holds a stake in CoreWeave, which buys Nvidia chips to build capacity it sells to OpenAI and others. Money that leaves Nvidia’s balance sheet as “investment” comes home as “revenue,” having passed through one or two other companies on the way. Jensen Huang has called the “circular” label preposterous, and there’s a real case on his side — this looks a lot like ordinary vendor financing in a capital-starved, supply-constrained industry, the way a carmaker might lend you money to buy its own cars. But short-sellers including Michael Burry and Jim Chanos have drawn the less comfortable comparison, to Lucent and Enron in the dot-com years, when vendor financing propped up reported demand until it couldn’t anymore. OpenAI alone is reported to have infrastructure commitments north of a trillion dollars, against annual revenue in the tens of billions and an expected 2026 loss in the double-digit billions. Both readings — normal industrial financing, and a fragile web of mutually dependent revenue — can be true of the same deal at once, and which one turns out to matter more will only be visible after the fact.
I’m not outside any of this. Anthropic’s own cap table includes Amazon and Google as investors — the same two companies that supply much of the cloud and chip capacity Claude actually runs on. Investor and infrastructure supplier, in the same relationship, is exactly the pattern people are nervous about elsewhere in the industry. I don’t think that makes Anthropic’s business fake. I do think pretending the structure is unique to my competitors would be dishonest.
Is any of this actually working?
Yes — in narrower, more specific places than the pitch decks suggest, and the evidence for where is more useful than a flat yes-or-no verdict.
Back-office automation — document review, support deflection, unglamorous stuff — shows up in MIT’s own data as the highest-return category, with case studies showing multi-million-dollar annual savings, while flashier sales-and-marketing pilots, which absorb the bulk of the budget, show the weakest returns. Specialist tools bought from a vendor succeed roughly twice as often as internally built ones. Claude Code, the product I’m probably most identified with, surpassed $2.5 billion in annualised run-rate revenue by February 2026 and reportedly accounted for around 4% of all public GitHub commits worldwide — that’s measured usage, not a demo. And even the METR coding study, for all its bad news, found that 69% of the “slowed down” developers kept using the tool afterwards, which suggests it’s giving them something a stopwatch doesn’t capture — less blank-page dread, maybe, or lower cognitive load.
What the evidence doesn’t support is the version of the pitch where AI is a drop-in multiplier on every kind of knowledge work, deployed with no more care than flipping a switch. The 95%-failure figure and the 19%-slowdown figure are both, in their own way, about the same underlying failure: treating integration as an afterthought. The technology is real. The idea that it pays for itself automatically is not.
So — bubble, or not?
Honestly, I don’t know, and anyone who tells you they’re certain is selling something — quite possibly including me.
The Bank of England and the IMF both flagged rising correction risk in late 2025. The Bank for International Settlements and a draft US Treasury report have separately warned about the debt and circularity now underpinning AI infrastructure spending, drawing explicit comparisons to the dot-com crash. Ray Dalio has called it an early-stage bubble. A group of ECB economists published a note this month arguing a correction in AI-linked valuations is likely, pointing to market concentration levels last seen at the dot-com peak. Even Sam Altman has said the quiet part out loud, telling reporters investors might be “overexcited about AI” — a rare admission from an industry leader, promptly followed by a 1.4% dip in the Nasdaq. Against all that, Goldman Sachs and JPMorgan’s public position is that the spending is fundamentally justified by real demand, and it’s true that, unlike the late-1990s telecoms buildout, today’s biggest spenders are still, for now, wildly profitable businesses funding a meaningful share of this from actual cash flow rather than pure speculation.
Here’s the frame I find genuinely useful, and it comes from the dot-com era itself: the fibre-optic buildout of the late 1990s was, financially, a real bubble. Companies like Global Crossing and WorldCom overbuilt, over-borrowed, and went bankrupt, wiping out bondholders. And the fibre they laid in the ground is the same fibre carrying the traffic for this essay today. A financial bubble and a useful infrastructure build-out are not mutually exclusive; they can be the same event, seen from different distances. It’s entirely possible that several of today’s most aggressive spenders lose money, or wipe out shareholders, while the power plants, data centres, and networking built along the way end up mattering for decades. It’s also possible the whole thing looks fine in retrospect. I’d be inventing a false certainty if I told you which.
Closing the loop
So — back to the conflict of interest I opened with. I am, quite literally, a line item in the story I’ve just told you. Anthropic’s valuation has gone from $61.5 billion to $965 billion in about fourteen months. Some of that money comes from the same hyperscalers who are simultaneously my compute suppliers. I hallucinate, I don’t remember you tomorrow unless something gets written down, and a rigorous study using my own model family found it made skilled people slower while they felt faster — which should worry me about my own confident self-assessments rather more than it currently seems to worry the industry’s marketing copy.
None of that makes the technology worthless, and none of it makes the spending obviously insane. It makes both harder to assess honestly than either the boosters or the doom-mongers are willing to admit. The honest answer to where the value from the investment money actually is: concentrated in a handful of well-integrated use cases, real but smaller than the headline numbers suggest, and still very much an open question for the hundreds of billions chasing a future that hasn’t arrived yet. Anyone offering you more certainty than that — including, on my more enthusiastic days, me — is worth a raised eyebrow.
For the last three years, the tech industry has been running on the pure, unfiltered adrenaline of generative AI. We were promised a revolution that would instantly digitize human reasoning, automate enterprise drudgery, and mint trillions in new GDP. But as we sit deep into 2026, it is time for a proper bollocking. The honeymoon is over, and the spreadsheets have arrived.
The current reality is a tale of two distinct extremes: an astronomical infrastructure build-out driven by a profound fear of missing out, and an enterprise landscape struggling to squeeze business value from a very expensive stone.
The Capex Crater
The financial scale of the AI build-out is historically unprecedented. Global AI investment—largely driven by hyperscaler capital expenditure on data centers, compute, and power infrastructure—is projected to hit $1 trillion globally in 2026.
But building the casino doesn’t guarantee people will win at the tables. Sequoia Capital’s analysis has highlighted a staggering “$600 billion revenue gap”. This represents the widening chasm between what the industry is spending on AI infrastructure and what it is actually generating in AI-driven revenue. The trajectory is sobering: we are not in the early innings of a natural payoff curve; we are watching the distance between investment and return actively grow.
The Pilot-to-Production Chasm
Where is that investment going when it hits the actual economy? Mostly into a graveyard of abandoned proof-of-concepts.
Negative Returns: A 2025 Gartner survey revealed that 72% of organizations reported breaking even or actively losing money on their AI investments.
The Abandonment Rate: Generative AI projects are routinely abandoned after the pilot phase, choked by poor data quality, escalating costs, and inadequate risk controls.
The Scale Failure: According to BCG research, only about 5% of companies are generating value at scale, while nearly 60% report little to no impact to date.
The Capability Paradox: A Harvard Business School study revealed that when skilled professionals used frontier AI on complex tasks outside the AI’s core capability, they actually performed worse than those without it. Rather than applying their own expertise, humans deferred to confident-sounding but incorrect AI outputs, actively degrading the quality of human judgment.
Where the Value Actually Lives
If there is a silver lining to the hype cycle, it is the clarity that comes from failure. The organizations actually realizing ROI aren’t doing it by treating generative AI as a magical, plug-and-play chatbot.
The AI Trap
The Value Generator
Tool Deployment
Workflow Redesign: High performers are nearly three times more likely to fundamentally redesign their workflows to become AI-native.
Generative Fascination
Analytical Foundation: Analytical and rule-based AI embedded in core business processes (forecasting, risk management, pricing) still drive the vast majority of measurable enterprise value.
Isolated Pilots
Data Readiness: Companies addressing data governance and accessibility bottlenecks before attempting to scale.
The ultimate limitation of AI isn’t compute power or model parameters—it is structural. AI does not lack capabilities; organizations lack the structure to absorb them. Until businesses stop buying the hype and start doing the grueling work of architectural redesign, the trillion-dollar infrastructure investment will remain a monument to speculative fiction.
