A shock on the scale of deindustrialisation is coming down the pipe for the United Kingdom. The leading indicators have already moved. Act now, with a Manhattan Project for the mind, a national programme to measure and train human capacity, and Britain does not just survive the AI decade. It becomes the world's first cognitive superpower.
Part one is the problem, built entirely on independent public data. Part two is the response. Every figure is sourced at the end of the page. Check all of it for yourself. Prefer it as a document? Open the briefing as a PDF.
Everything in this part is from independent public sources. Check all of it without us in the room.
UK job adverts for programmers, 2019/20 to 2024/25
Adverts for programmers in the United Kingdom have fallen by close to 70% in five years, with the decline accelerating in the last two. No recession. No factory closure. No banking crisis. AI became cheaper than a graduate.
Four senior programming roles are now advertised for every junior one. The bottom rung of the ladder is being removed first.
2024 saw the first fall in the number of working software developers since 2006.
This is the most prestigious entry level cognitive job in Britain, and it is disappearing.
Source: NFER / The HG Foundation, Destination Tech, 2025.
It is not just programmers. Across the labour market, hiring is falling twice as fast where AI exposure is highest.
The decline concentrates in software, data, design, media, research, legal, HR, finance and business roles.
Up to 3 million UK jobs in declining occupations could disappear by 2035.
Job adverts are the earliest moving labour market signal. They move years before the unemployment statistics.
Sources: McKinsey UK, 2025; NFER, The Skills Imperative 2035 (Nuffield Foundation), 2025.
Working age people economically inactive through long term sickness
The system expected to absorb the income shock, the status shock and the purpose shock of cognitive automation is already at saturation, before the new load arrives.
1.35 million of those inactive through sickness report depression, anxiety or stress.
25.8% of 16 to 24 year olds have a common mental health condition, up from 17.5% in 2007.
The figures are the highest on record, and the young are the worst affected.
Sources: ONS economic inactivity series, 2024; NHS England, Adult Psychiatric Morbidity Survey 2023/24.
16 to 24 year olds not in education, employment or training
957,000 young people are NEET. Entry level cognitive work is how every cohort since 1945 converted education into income, and that conversion mechanism is the thing now being automated.
A cohort that cannot get its first job does not form households, does not buy houses, and does not pay the taxes the state is counting on.
An economy that stops hiring the young is borrowing against its own future tax base.
Source: ONS, young people not in education, employment or training, Q4 2025.
Working age health related benefit spending is rising fast, and the state has no plan to stop it.
Mental health is the dominant driver of the rise.
The £66bn forecast assumes the labour market carries on broadly as it is. The sections above say it is not.
Source: OBR forecasts as analysed by the Institute for Fiscal Studies, 2025.
Of all income tax is paid by 10% of taxpayers
That 10% are drawn overwhelmingly from cognitive professional work, which is the work being automated. If their incomes compress, the tax base compresses with them, at the moment the welfare line is rising.
The top 1% alone pay 28.2% of all income tax. HMRC, 2024-25.
Public sector net debt stands at 95.6% of GDP, levels last seen in the early 1960s.
The cohort that funds the state is the cohort most exposed to the shock.
Sources: HMRC income tax liabilities statistics, 2024-25; ONS public sector finances, November 2025.
Children born per woman. The lowest on record, third year running.
There is no demographic cushion under any of this. Total fertility in England and Wales is a third below replacement level, and family formation tracks economic confidence among the young, the cohort losing labour market access first.
Fewer workers, more dependants, a smaller tax base. Each line of this briefing compounds the others.
Source: ONS, births in England and Wales, 2024.
The Cognitive Shock models a central scenario of a 20 to 30% reduction in cognitive professional employment by 2035, built from the hiring data, the adoption curves and the cost arithmetic of AI substitution. One shock, cascading through six systems.
A point of honesty that matters: the scenario is the author's, modelled, and presented as such. The leading indicators in sections 01 to 07 are not modelled. They have already happened.
Source: The Cognitive Shock, chapters 2 to 9.
An orderly response is possible in the next two to three years.
The hiring data has already moved, and job adverts lead unemployment statistics by years. Waiting for the unemployment data means starting late by definition.
The welfare line compounds. Every cohort that enters inactivity adds a recurring annual cost.
The response takes years to build. Infrastructure at population scale has to exist before the peak load arrives.
Nothing here requires the worst case to be true. The observed data alone justifies preparation.
The shock is not optional. Displacement is. Measure. Train. Recognise.
The basis of the response is the set of capacities AI does not replicate. In an AI saturated economy, these are not soft skills. They are what humans will be uniquely paid for.
Holding focus on what matters, in an environment engineered to fragment it.
Judgement when the data runs out. The models still hand this back.
Staying functional when the stakes are real and the clock is running.
How fast a person acquires what the economy newly values.
Knowing who you are when the job title no longer answers the question.
Acting on the world rather than being processed by it.
Source: The Cognitive Shock, chapter 10: What AI Cannot Do.
A Public Health England for human capacity.
What is not measured at scale is not managed at scale. The state already measures attainment, income, health outcomes, fertility and mortality at population level. It measures none of the six capacities above. The proposal is measurement of attention, regulation, learning velocity, decision quality, identity coherence and agency at the level of the individual, the team, the institution and the population, with a common methodology and proper privacy and consent protections.
Voice is the most efficient measurement substrate. A five to ten minute structured voice session, repeated periodically, captures the major signal across five of the six capacities.
The cost per person per year is a few pounds at current AI inference prices.
Technically achievable today. The constraint is institutional, not technical.
Source: The Cognitive Shock, chapter 11: The Infrastructure.
The methods exist. The deployment does not.
Measurement finds the gap. Training closes it. The evidence base is not speculative: millennia of contemplative practice, fifty years of clinical CBT literature, and thirty years of deliberate practice research. What is missing is integration, funding, and the political coalition to deploy it through three channels.
Schools: explicit curriculum time for attention, regulation, learning and identity, from primary to sixth form.
The workplace: employer programmes that build capacity inside daily working life, not as a separate wellness activity.
Civic life: rebuilding the institutions that historically carried meaning, community and practice.
Source: The Cognitive Shock, chapter 11: The Infrastructure.
From credentials to measured capacity.
British institutions reward credentials: GCSEs, A levels, degrees, job titles. Credentials are proxies for capacity, and they are calibrated against precisely the routine cognitive work AI now substitutes. The shift required is to recognise and price measured capacity directly.
A hiring manager in 2030 weighs a first class degree against a recent capacity assessment showing high attention, regulation, learning velocity and decision quality, and hires the second, because measured capacity is the better predictor of performance.
This is already visible at the leading edge of US technology hiring. Whoever builds the recognised measure sets the standard. Britain can set it.
Source: The Cognitive Shock, chapter 11: The Infrastructure.
£1bn to £3bn a year over a decade for the measurement and training infrastructure. Between 0.1% and 0.25% of total government spending. A rounding error against the budgets it protects.
The analogues are electrification, the road network and the NHS itself. Each was built in a generation, at meaningful but achievable cost, with returns compounding across every sector that depended on it.
The cost is not the constraint. The political will is.
Source: The Cognitive Shock, chapter 11.
The first pieces of this infrastructure are already being built in Britain, by two companies in the same founder lineage. Interest declared: the author is the CEO of TrueMind, and Will Mellors-Blair, who wrote the book's foreword, founded TrueMind and leads Mindora. Everything in part one stands independently of both.
The elite performance play. Voice based measurement of the six capacities in elite environments, starting in football academies: high stakes, duty of care, clear performance return.
The pathway beyond football is the infrastructure architecture itself: industry, defence, education, the state workforce response.
The broad workforce play. Biometric and behavioural measurement across the whole workforce: sensors, sleep, mood and movement, owned by the individual.
Backed by the UK Government's Global Entrepreneur Programme at the Department for Business and Trade.
Two plays at one category, and both early builders are British. No country owns this category yet.
Source: The Cognitive Shock, chapter 12: What We Are Building.
Three asks, in ascending order of commitment.
Plan on the leading indicators, not the lagging ones.
A cross departmental assessment of UK exposure to cognitive automation: labour market, welfare, education and tax base.
A small measurement and training pilot in one public cohort, schools, retraining or the NHS workforce, with published results.
For a conversation, get in touch.
Every figure in part one is from an independent public source.