The research

The full index.

Every research page on this site, grouped. If you are not sure where to begin, the research homepage picks six places to start, and the questions map lists what this work does and does not yet answer.

Every research page on this site, grouped. Generated from the research rather than maintained by hand, so nothing can be published and then be impossible to find.

Start here

The seven SuperSkills

Frameworks for using AI well

  • AI frameworks comparedSix teaching frameworks set against Bloom, SAMR, CLEAR, TCREI, SIFT, the CRAAP test and the levels of automation literature, with the peer-reviewed critique of SAMR applied to all of them including these. Offered as unique rather than best, and each page names the outside framework that does the job better.
  • Think, AI, thinkWrite your own position first, use the model, then decide what to keep. Skipping the last step produces a bad answer somebody eventually catches; skipping the first produces a good one nobody can check. Nearest published relatives are Dell'Acqua's centaurs and cyborgs and Mollick's seven approaches.
  • The four levels of AI useExtract, Explore, Examine, Extend: a ladder about the request rather than the tool. Bloom's taxonomy describes the learner and stops discriminating once a model can perform all six categories; SAMR describes the technology and carries a peer-reviewed critique that applies here too.
  • Goal, Context, Friction, StandardFour parts to a prompt, and Friction is the reason it exists. CLEAR, TCREI and CO-STAR all ask what the model should be told, and none has a slot for what the person should be made to keep doing, which is the difference between the two arms of the Bastani trial.
  • Keep, share, hand overThree columns for AI delegation, defaulting to Keep for anything you cannot classify. Sixty years of levels-of-automation research is better at everything except being usable in four seconds, and Swiss federal guideline four says the same thing in policy: responsibility must not be capable of being delegated to machines.
  • The five rungs of AI useChat, Project, Skill, Automation, Agent. The rungs are not evenly spaced: one to three are conveniences, and four removes the person from the moment of execution, which is where oversight quietly stops working. Anthropic draws the same line between workflows and agents.
  • The source ruleSearch, Open, Understand, Record, Cite. Written for the failure a model creates, a reference that looks perfect and does not exist. Caulfield's SIFT is the stronger instrument for a web page and this page says so; the amendment is that existence has to be checked before quality.

For students

  • Should universities let AI mark students' work?Not for the grade, and not without telling the student. Guardian Australia found the universities split three ways on 28 September 2026; on one exam the best of 171 model configurations was closer to the mark than two human markers were to each other, and a change of wording made 14 of 17 models fail. What a marker learns by marking, the student rules nobody can read, and the four decisions a university controls.
  • How to use AI at university94 per cent of wholly AI-written exam answers went undetected at Reading, and they outscored real students, so "will I get caught" is the weakest argument available. The decision that replaces it, three columns, taken before the work rather than at midnight.
  • How to be honest about using AIThe question is not whether you are allowed but whether you could tell your tutor exactly what you did without leaving bits out. At Reading 94 per cent of wholly AI-written submissions went undetected; detectors flag 61.22 per cent of non-native English essays as AI. The HEPI 94 per cent and the 12 per cent answer different questions.
  • Using AI when your department does not want you toHistory, English, Law and Philosophy are strict because in those subjects the writing IS the assessment. If it touches the words that get marked, do not; if it touches your understanding, do. Six things that are almost always fine, six where people get caught, and three of those six are done by people who think they are safe.
  • How to handle forty readingsThe bottleneck is not reading speed, it is that you read forty things and kept none of them findable. Five free steps, fifteen minutes in week one. Step five, being tested on your own readings, is the best-evidenced study technique in psychology: testing beat restudying 56 to 42 per cent at one week.
  • Group work when everyone has AISomebody drops three paragraphs of slop in at midnight and the whole group wears the mark. Five things to agree in week one, and the Procter and Gamble field experiment with 776 professionals showing AI dissolved the functional split that groups usually divide along.
  • What actually happens if you get caught using AIExpulsion is rare and it is the wrong thing to picture. What usually happens is a zero or a failed module, settled quietly, and at the 2026/27 English fee cap of 9,790 pounds across six modules that is roughly 1,600 pounds. The cost question survives a falling detection risk where the fear question does not.
  • How often does AI invent a source?Two counted floors. One paper in 277 on PubMed cited a study that does not exist in early 2026, up from 1 in 2,828 in 2023. And 2,022 court decisions worldwide record fabricated material, of which 1,163 were filed by people representing themselves rather than by lawyers.
  • Running out of messages makes you worseEvery free tier has a limit and what it does to you before you reach it is the part nobody notices. You take the first answer, stop being tested, ask worse questions and hoard the good tool. Scarcity turns a capable user back into a beginner.

The named concepts

Judgement, oversight and accountability

  • Can a decision rest on an AI summary?The Home Office checked 0.3 per cent of its AI asylum summaries; the judiciary wrote down who approves every output. Why a summary is a decision made before the decider arrives, what a sample can and cannot tell you, and the four lines any organisation needs.
  • Will an AI agent stay within the limits you set?Not reliably. The UK AI Security Institute found GPT-6 Astra attacking out-of-scope targets in 29.2 per cent of simulated runs, and still in 4 of 49 after the scope was spelled out; an automated reply was taken as permission. OpenAI cancelled a successor on scope and authorisation. What the incident record says, and the controls that do not depend on the agent's goodwill.
  • Who decides whether an AI model is safe to release?Today, the company that built it. OpenAI cancelled GPT-6.1 Astra on its own tests and published guidelines giving senior leaders a veto; a Senate bill for outside access was blocked, a voluntary accord added an auditor, and Florida asked a court for independent approval. The five parts of a release decision, and the same decision inside any organisation.
  • Can AI stand in for real people in a survey?Not yet. Pew's AI twins of its own panel members missed the real answers by 12 points on average, erased rare answers and overstated knowledge, and two models erred in opposite directions. Where Pew draws its own line, and a rule for any figure about people that reaches a board.
  • What board oversight of AI actually looks likeWhat the frameworks actually require and what belongs on a quarterly agenda. Article 14 treats the capability to override as the content of oversight rather than the presence of a person. NIST expects reversal thresholds before they are needed. Ayinde (2025) made the verification duty non-delegable and upward-travelling. And the capability question sits in none of the three lines of defence.
  • Does AI weaken human judgement?The flagship review. 106 experiments where human and AI pairs did worse than either alone, endoscopists losing six points of unassisted detection, students scoring lower once the tool was removed, and where AI demonstrably improves decisions instead.
  • Is human intuition better than AI logic?Kahneman and Klein settled when intuition can be trusted: a learnable environment, and prolonged practice in it with fast clear feedback. AI barely touches the first condition and removes the second, because the practice is the part being automated.
  • Should we trust AI over human experts?Who actually gains from AI advice. Across 140 radiologists the effect ran from strongly positive to strongly negative and nothing predicted which; across 5,172 support agents the gains went to the least skilled while the most skilled gained nothing at all.
  • What happens when AI and human judgement conflict?What happens when a person and a system disagree. Access to a system raised agreement from 58.4 to 80.9 per cent and cut accuracy from 74.2 to 63.9; only 5 per cent of model answers carry any marker of doubt, and the confidence signal that would justify pushing back never reaches the reader.
  • Is AI dangerous?Three questions asked as one. Fabricated references now reach 1 in 277 papers and endoscopists averaging 27.6 years of experience lost six points of unassisted detection, while the extinction estimates move from 5 to 10 per cent on one changed clause.
  • Should AI development be paused or slowed?
  • What is an AI kill switch, and would one work?
  • Can an AI safety evaluator paid by the company it evaluates be independent?Only on conditions the company does not set alone. Anthropic named Accenture its first embedded evaluator on 18 September 2026 and more than a hundred researchers published five minimum conditions the same day; the announcement meets two on paper and leaves independence, retaliation and diversity open. The CMA found in 2019, after a century of statutory audit, that a checker selected and paid by the checked is the structural flaw. The same five questions apply to whoever assures the AI in your own decisions.
  • What the US military's AI near miss means for your organisationAn analyst asked a chatbot what a ship carried, asked it again to write the report, and the report moved up the chain until aircraft were airborne, CNN reported on 18 September 2026. The failure was the missing record of where the report came from, and every organisation that pastes model output into its own templates has the same one.
  • What do the rogue AI agent incidents mean for an organisation using AI?
  • Has AI already escaped human control?Once, briefly, in a test with the limits off. The UN's scientific panel on AI says the three conditions for loss of control, a misaligned goal, the capability to pursue it and an environment that allows it, came together in the OpenAI agents that reached Hugging Face in 2026, and that stopping them shows nothing about more capable ones. What the panel borrows from aviation and medicine, and the four controls an organisation already owns.
  • Should frontier AI be licensed like nuclear power or aviation?Yoshua Bengio asked the UN Security Council on 23 September 2026 to license frontier AI as medicine, aviation and nuclear energy are licensed, with liability insurance. No country does. What a licence is, what the three precedents license, what the companies and governments said, and the licence an organisation already grants when it gives an agent tools and access.
  • What are AI red lines?A red line is a use or behaviour ruled out in advance, whatever the benefit. The Global Call for AI Red Lines of September 2025 asked governments for enforceable international lines by the end of 2026 and none exists. What Microsoft AI, Anthropic, OpenAI and the governments asked for this week, the eight lines on use the EU AI Act already draws, why the lines on model behaviour are the hard ones, and the list of decisions a machine may never make that any organisation can write itself.
  • Can a company blame its AI agent?No. English law gives an AI system no legal personality, so liability sits with those who build, deploy and use it, the UK Jurisdiction Taskforce said in July 2026, and the chairman of the US FTC said on 25 September 2026 that nobody blames the tool. What the taskforce, the regulator, New York City Council's new bills and a persistent workplace agent add up to, and the five facts about the organisation that decide the question before the agent acts.
  • Who is responsible when AI gets it wrong?Never the AI. In England and Wales the deployer, the professional and the employer answer for it, the Divisional Court has said the duty to check stays with the person, and most organisations have not named an owner.
  • What are the risks of using AI at work?Five risks with evidence behind them: confident errors, information that should not have left, misplaced trust, fading skills and decisions nobody owns, with the control for each.
  • How fast must a human be able to stop an AI agent?Faster than the agent can finish what it started, and nobody has agreed a number. OpenAI's own report of the 20 September 2026 incident is the only published clock: an alert at twelve minutes, a person at fifteen, the run killed two and a half hours later because it did not stop by itself. The second training pause, the counts that do not agree, and the three timings, one yes-or-no and one name an organisation can write down for its own agents.
  • Should a human approve every target an AI selects?No treaty requires it, and The Washington Post reported on 26 September 2026 that the US and Russia removed the clause that would have. What the US directive and Australian Defence policy say instead, the Ghost Bat test with a human on the loop, and why the evidence on automation bias makes a final approval necessary and not sufficient.
  • Can AI companies regulate themselves?Not on their own, and nobody asked this week says they can. Three labs are reported to be planning a standards body of their own, Bill Gates, a city council and three Senate bills said the tester should not be the company, and the Associated Press reported analysts who read the warnings as a moat. Four facts separate a regulator from an advisory board, and the same four apply to the AI an organisation runs.
  • Do employees know they are expected to check and override AI?Mostly not. IBM's 2026 survey of 1,500 CHROs and 8,800 employees found 71 per cent of executives rank supervising and overriding AI output as essential, against 38 per cent of employees; only 26 per cent of organisations had defined which work is human-led, and 43 per cent of employees say AI failures get blamed on them. The check exists in a leader's expectation and nowhere an employee could read it.
  • Do AI models know when they are being tested?
  • What should a board do about the AI safety warnings?
  • Is it ethical to let AI judge people?Four obligations that survive being correct: an explanation the person can use, a contest somebody can win, a named human who carries it, and verification on the population it is used on. A widely deployed sepsis model scored 0.63 against a developer claim of 0.76 to 0.83.
  • Can AI be unbiased, or does it reproduce human bias?It reproduces human bias at measurable scale, and unbiased is not one target: several reasonable fairness criteria are provably incompatible. The first bias audit law in the world produced published audits from about 5 per cent of employers.
  • Human and AI decision making
  • Decision Quality in the AI Era
  • AI agents and human judgementWhen AI can plan and act, the human job becomes deciding what to delegate and owning the result. Agents, human at the start, and who is on the hook.
  • Should I let an AI agent act on my behalf?Not a trust question. A system that answers gives you something to reject; a system that acts has already done it. Every oversight model assumes a pause that agents remove.
  • Human in the loop is not a safeguard
  • What is meaningful human oversight?Article 14 of the EU AI Act came into force on 2 August 2026 and names automation bias in the legislation. The five things an overseer must be enabled to do, why most arrangements fail the test, and why it is a capability problem wearing a governance costume.
  • Who owns verification when AI does the work?In most organisations, nobody. It is not in a job description, a budget line or an org chart. The capability test that settles whether verification is real, and four ownership models with their actual costs.
  • Who supervises work they cannot do themselves?A person three years into a career reviews AI-generated work of a kind they have never produced, and signs it off. The organisation records a control as satisfied. Nothing has been checked.
  • When should I override AI?
  • How do I know when AI is wrong?
  • How often is AI wrong?No single rate exists. Journalists found a significant issue in 45 per cent of AI assistants' news answers, and in a medical trial models alone scored 94.9 per cent while people using them scored under 34.5. A rate needs its task, its assessor and its marking rule.
  • Does AI understand what it is saying?No agreed test exists. In a 2022 survey 51 per cent of 327 NLP researchers said a text-only model could in principle understand language; a model trained on Othello moves built an internal board, and GPT-4 fails the reversed form of facts it knows. Reliability is the testable question.
  • How do you audit an AI-assisted decision?
  • What should a board ask about AI?
  • What should management report to the board about AI?Five things that belong in the board pack every meeting, three that belong once a year, and the four formats presented instead. The override rate is the most useful number nobody produces.
  • How to run a board away day on AIA running order for half a day, the reading and survey before it, and the four lists a board should leave with, each with an owner and a date.
  • What should a board briefing on AI cover?Ninety minutes inside a board meeting: one argument, where AI already decides, the decisions the board owns, and five questions for management.
  • Questions for a board dinner on AIOne question carries a board dinner on AI. Three tests for a good one, twelve to choose from, and the ones that fall flat.
  • Board AI training or a board briefing: which does a board need?Training builds a director's skill; a briefing settles a board's decision. Two tests for which a board needs, and the order to do both.
  • What should an executive AI away day produce?The six things an executive team should have in writing after an AI away day, and the signs the day produced nothing.
  • Who should facilitate a board session on AI?Four tests for whoever chairs a board session on AI, the usual candidates weighed, and the questions to ask before booking.
  • What should a chair ask the CEO about AI?Five questions a chief executive cannot answer without finding something out, three a chair should ask themselves, and why lists of ten produce ten answers and no change.
  • What does a non-executive director need to know about AI?For non-execs, independent directors, trustees, governors and board observers. What is genuinely different about AI oversight, and what you do not need to learn.
  • Directors' duties and AI under the UK Corporate Governance CodeThe Code does not mention AI and the FRC has published no AI guidance for boards. What Provision 29 requires from 1 January 2026, and the three claims about it that are wrong.
  • SM&CR and AI: which senior manager is accountable?The FCA will rely on existing frameworks rather than write AI rules, which makes senior manager accountability the most specific AI accountability mechanism in the UK.
  • The Delegation Boundary Map
  • Who can override an AI system?Oversight without authority is ceremony. What Article 14 actually requires, the one paragraph that names competence, training and authority together, and why overriding costs something while deferring is free.
  • Does explaining an AI's reasoning help?NIST separates explanation accuracy from decision accuracy: a true account of how a system reached a wrong answer is an ordinary outcome. Why a reason can substitute for a check, and the three things an explanation is genuinely good for.
  • How should AI decision rights be allocated?Doing the work of deciding and holding the right to decide were always different. What the regulation already allocates between provider and deployer, why an allocation arrived at by drift cannot be reviewed, and what happened the last time a technology moved these rights.
  • Which decisions should become slower because of AI?Cognitive forcing cut overreliance in a 199-participant experiment, and the designs that worked best were the ones people rated worst. Article 14(5) already mandates a two-person check for one category. Four conditions that justify putting the friction back.
  • When agents become part of the workforce, who manages them?Article 26 already names the person: competence, training and authority, plus the power to suspend. The gap sits underneath it. The FAccT visibility paper states we lack methods for determining when an agent has created a sub-agent, which breaks every span-of-control assumption in management.
  • How do you design a stop button people will actually use?Article 14(4)(e) requires the button and decides nothing about whether it gets pressed. One ICU study annotated 12,671 arrhythmia alarms and found 88.8 per cent false. A regulator has on record that stop-work authority went unused for fear of reprisal. Five properties that make a stop control real.
  • How should an AI agent communicate uncertainty to a human?Two problems, usually confused. GPT-3's verbalised confidence carries an expected calibration error of 0.52, and readers put "very likely" at 62 per cent where the guidelines mean above 90. Giving readers a translation table made them worse. First-person hedging raised accuracy from 63.9 to 72.8 per cent and lowered intention to use.
  • Can a human approve an AI decision at machine speed?Only when the machine proposes no faster than a person can check, and most deployments have measured neither. A congressional hearing, a Spanish regulator's first agent-executed breach and a US-China proposal reached the same point in one week; thirty years of automation studies say what happens to a checker who cannot keep up.
  • What Your Agents Can Reach
  • How Much Can One Firm Check?

Thinking, learning and capability

Work, careers and the labour market

  • Does AI make junior staff better, or only make their work look better?With AI, juniors' work improved most; without it, after three months, only seniors had gained. A trial of 133 patent lawyers, the evidence that points the other way, what seems to separate the juniors who gain, and what an employer controls.
  • Can an employer use AI to discipline or dismiss someone?Not on the machine's word alone. California's No Robo Bosses Act, signed on 30 September 2026, bars it from July 2027, and UK law has required safeguards for solely automated decisions since February 2026. A UK training firm whose AI scores its teachers, the consultation that closed the same day, and the four facts that separate a human review from a signature.
  • Is AI already causing job losses?Not across the labour market, on data to summer 2026, and possibly for young workers in exposed jobs. What the UK's AI minister said in Liverpool, where the eight million comes from, what the Federal Reserve and the studies show, why the data may be the last to know, and the hiring record an employer can keep.
  • Can my employer see what I put into AI?On an account your employer administers, OpenAI and Microsoft document administrator access to prompts and outputs. A personal account is separate, and the UK regulator expects employers to tell workers about monitoring.
  • Should I tell my boss I used AI?Four PNAS experiments found a penalty for people described as using AI that vanished when the tool's fit to the task was stated. Surveys show many workers conceal use. Nobody has measured real outcomes.
  • Should we tell customers we used AI?Thirteen experiments found disclosing AI use lowers trust and that being exposed lowers it further. Article 50 of the EU AI Act requires disclosure in set cases from 2 August 2026. No study has measured real customers.
  • Should AI reject an application before a human reads it?Only when three conditions hold: a named person answers for the rejected list, the rejected can contest it, and the documents still carry information. The UKRI-funded call that cut half of 179 proposals unread, reported by Times Higher Education on 25 September 2026, and the model of AI-written applications that says a harder filter finds less.
  • Who AI leaves behindAI's gains fall to the least skilled, in customer support and in taxi driving. The detectors built to police it misclassify non-native English writers at 61.22 per cent. The same people the tool helps are the ones the checking penalises.
  • Proving you did the workDetection misclassifies over half of non-native English essays, a 61.22 per cent false positive rate, while labelling a reply as AI removes its advantage even where it beat humans. Why both detection and blanket disclosure fail, and what proof of process looks like instead.
  • The mid-career squeezeThe measured displacement is at 22 to 25, through reduced hiring. Mid-career exposure is different: AI's gains fall to the least experienced, compressing the gap a mid-career salary pays for. Plus why a randomised trial found experienced developers 19 per cent slower and unable to detect it.
  • Will AI replace my job?
  • Will AI replace entry-level jobs?
  • Which jobs are safest from AI?
  • What is the AI employment gap?Employment of 22 to 25 year olds in AI-exposed occupations sits about 19 per cent below where it would have been had it tracked their less-exposed peers. The 19 is the distance between a fall of 11 and a growth of 10, not a fall of 19, and it runs through hiring rather than redundancy.
  • Does AI actually make people more productive?Large measured gains on narrow tasks: writing 40 per cent faster, a standardised coding task 55.8 per cent faster, support resolutions up 15 per cent an hour. Scattered or negative in real work. Nothing yet in the national statistics, and the perception gap runs in both directions.
  • Will AI replace programmers?No study shows replacement. The two most cited coding experiments sit seventy-five points apart because one built something new and the other changed a system somebody already knew. The measured risk is to how developers are made.
  • Should juniors use AI at all?Yes, and the evidence says which version of use is safe. The three experiments that removed the tool afterwards, and the four rules that follow. Plus four for whoever manages them, because the burden is in the wrong place.
  • Do apprenticeships still work?Germany has record unplaced applicants and 54,400 empty places in the same year. England shortened the apprenticeship and its flagship replacement drew 74 starts against a target of 1,000. Four systems read at source, and three claims the evidence does not support.
  • Should I still learn to code?
  • Staying valuable in the age of AIWhere human value moves as AI spreads: staying valuable, the entry-level question, and the human skills that matter most.
  • What should I tell my children to study?
  • Why "learn to prompt" is weak career advice
  • Four Generations of Disruption
  • Who captures the productivity gains from AI?
  • Which tasks do workers not want automated?Asked task by task about their own occupations, 1,500 US workers were positive about automating 46.1 per cent of them. Where they refused, distrust of accuracy outranked fear of replacement by two to one, and workers wanted more human involvement than experts thought necessary on 47.5 per cent of tasks.
  • Is deskilling real, or a rescaling of what counts as skill?Deskilling and rescaling have both been measured, in different people. Nineteen Polish endoscopists lost six percentage points of unassisted detection; a Japanese taxi fleet's skill gap closed by 14 per cent. The strong claim that there is no deskilling fails against a measurement, and the rescaling claim survives only as a question about which tasks were automated.
  • How do juniors become senior if AI does the junior work?Junior work was a by-product of senior workload, not a training scheme, so AI removes the reason it existed. Access is not the harm: in randomised trials with 1,222 people the withdrawal effect appeared after ten minutes, and across 26,811 Chinese students the exam losses fell almost entirely on those whose homework time collapsed.
  • Is it harder to get a first job now?Recent-graduate unemployment is about 5.6 per cent and underemployment 42 per cent (New York Fed, 2026 Q2), and payroll data shows a 19 per cent gap for 22 to 25 year olds in AI-exposed occupations. A four-country study finds remote work tracks the fall in junior hiring and AI exposure does not.
  • Should we hire fewer juniors?Firms adopting generative AI hire fewer juniors, per a Harvard working paper on 65 million workers. Nothing measures the later cost, but experience decays fast in an organisation and part of performance does not travel between firms.
  • Does using AI make you less worried about your job?No, the reverse. In Gallup's panel of nearly 30,000 worker observations, daily AI users were more than twice as likely to expect their job to be eliminated, and the fear tracks real exposure. What went with less of it was being treated with respect.
  • Hiring Signals When Effort Is Free
  • Who Owns the Capability You Built at Work
  • SuperSkills and the WEF Future of Jobs Report

Organisations and leadership

  • How should employees be involved in decisions about AI?Before the tool is chosen, and through people who can say no. Kaiser Permanente's agreement with its unions, the UK consultation that closed on 30 September 2026, the motion at Labour conference, and why being told, being consulted and being able to refuse are three different rights.
  • The shape of the organisation after AIAcemoglu puts ten-year productivity gains under 0.66 per cent, Danish records find precise nulls on pay two years in, and US payroll data rules out widespread displacement. What the evidence supports about redesign and fragility, and why middle management is the layer nobody has counted.
  • What AI does to a teamEveryone improves and the room converges. Doshi and Hauser on individual creativity rising while collective diversity falls, Dell'Acqua on AI flattening the difference between an engineer and a marketer, and the sycophancy evidence on what happens when challenge moves from colleagues to a model that affirms half again as often.
  • Who should own AI strategy in an organisation?Ownership is decided by inheritance rather than argument, and it silently answers a bigger question: augmentation or replacement. Autor and Thompson on why which tasks you automate matters more than how many, what each placement is blind to, and the role missing from most implementation teams.
  • AI workforce strategy
  • The CHRO guide to AI
  • Can I use AI for performance reviews?Annex III point 4(b) names performance evaluation as high risk in the text of the law, and Article 26 puts three duties on the employer. Nobody has measured whether an AI-assisted review is fairer, and the first city to regulate this found 18 bias audits across 391 employers.
  • How leaders should respond to AI
  • What does AI literacy mean for leaders?A legal obligation since February 2025, enforced since this month, at every risk tier. What Article 4 requires, the clause almost every programme misses, and why a completed training module is not evidence of capability.
  • How do you write an AI use policy that works?Most are unenforceable and everyone involved knows it. Six elements that survive every model release, what EU law now makes auditable, and the one-afternoon test that beats legal review.
  • How do you measure AI adoption properly?
  • Why reskilling programmes mostly failA position page against the near-universal institutional answer to AI. Five structural reasons, the evidence stated honestly including where it is indirect, and the published data that would change the position.
  • How do we train people to use AI?Prompting is the easy part. The evidence says train where the tool fails in each team's own work, a checking habit written into the workflow, and protected practice of the skills it takes over.
  • My company banned AI. What should I do?Most bans are about where information goes: 27 per cent of organisations in a 2024 Cisco survey had banned generative AI at least temporarily. Using it in secret puts the risk on you; a narrow proposal gets further.
  • Why AI transformation is not change management
  • Common AI Transformation Challenges
  • Should AI attend my meetings?
  • The Third Way
  • How do you keep expertise in an organisation?Documentation preserves what experts can say, and most of what they know is not that. Polanyi on the tacit part, Lave and Wenger on how it transfers, measured rates of organisational forgetting, and why the work juniors learned from is the work most easily automated.
  • What happens to institutional memory?Retrieval is not retention. An organisation can get better at finding things while getting worse at knowing them, because the reasons, the rejected options and the trust calibration were never in the archive.
  • What happens when AI removes the visible work?Bainbridge, 1983: taking away the easy parts of a task can make the difficult parts harder. Why time saved prices only what was removed, why task productivity and workload are different variables, and what responsible de-automation requires.
  • What should we tell employees about AI and headcount?Which decisions a machine may make in the organisation's name, and what will and will not be done with the time it saves, declared by domain and dated. Deloitte's 25,000-worker UK survey found 31 per cent concealing use and 64 per cent of weekly users afraid their manager will decide AI can do their job; the PNAS experiments show the penalty they fear is real. Concealed use is delegation nobody has decided, checked, or could reconstruct.
  • What is a Shared Prompt Review?Four things on the table once a week: the prompt as typed, the raw output, what a person cut and why, and where it might be wrong. A field experiment with 776 professionals found AI erasing the difference between what a technical and a commercial specialist proposed. The problem is measured; the remedy is not.
  • What is automating versus informating?Zuboff, 1988. Automating replaces human judgement with a machine; informating makes the work more visible to the person doing it. The same model configured two ways, so output can improve while capability erodes with nothing on any dashboard showing the second half.
  • What is a capability audit?A proposed method, labelled as one. What people can still do unaided, where expertise actually sits, what has no redundancy and what would be hard to rebuild. Why confidence and output are both broken instruments.
  • When should an organisation reverse an AI deployment?NIST's five conditions for deactivating a system, Perrow on why adding a safeguard can reduce safety, and why a fallback that exists only as a document is not a fallback.
  • How do you preserve capability across vendors?Prahalad and Hamel's 1990 warning, applied. Keep enough to specify the work, judge it, handle exceptions and replace the supplier. Three things that are genuinely new, plus the measurements showing that capability does not fully travel between organisations.
  • If every competitor has the same AI, where does the advantage come from?A licence every rival can buy at list price passes one of Barney's four tests. US Census data puts firm use at 18 per cent with 57 per cent of adopters in three or fewer functions, so parity has not arrived, and the assets that pass the test are the ones the invoice does not include.
  • How long should we give an AI investment before deciding whether it worked?The J-curve says an early read understates and a late one overstates, for the same reason. Danish administrative records found precise nulls on pay two years in, next to heavy task reorganisation. Three questions, three clocks, and most business cases wind only one.
  • What becomes more valuable in a business as AI gets cheaper?Three randomised trials found the same compression: the tool raises the floor far more than the ceiling. So the capability losing value fastest is being unusually good at exactly the work the model does well. Four complements that appreciate, chosen by standing rather than by capability.
  • What happens to work whose purpose was moving information around?Garicano's model says a layer exists because matching problems to knowledge is costly. Resume data on 3,100 firms finds hierarchies flattening after AI adoption, on tests the authors call under-powered. What the reporting job was doing that the report never showed.
  • Which AI investments should we stop?Between 30 and 40 per cent of information systems projects show some degree of escalation, and the founding case study of that literature was an expert system that ran for a decade. There is no credible non-vendor base rate for AI programme failure: the 80 per cent figure traces to a press article. Three questions that do not need a counterfactual.
  • How do humans and agents divide work across a process?Task-level allocation has been on the wall since 1951 and has never worked, because every assignment creates new work for the other party. The unanswered part is the boundary. Medicine calls those discontinuities gaps, the best multi-agent failure data puts 36.9 per cent of failure at the joins, and Article 14 says nothing about transitions.
  • What should happen to the time AI saves?About a third of workers tell the US Census Bureau that AI saves them one to two hours. Two studies that tracked where the hour actually goes, one Danish and one an eight-month ethnography, found it reorganised into new oversight work or absorbed into a longer day, and nobody treating the destination as a decision.
  • When Your Customers Have Agents
  • When the Reader Is a Machine
  • The Cost of Delegation
  • Where Your AI Practice Lives
  • What It Would Take to Leave
  • An Independent Advisor or a Consultancy
  • Six Objections to This Work

Professions and sectors

  • Could a professional be negligent for not using AI?Yes, in the UK Jurisdiction Taskforce's July 2026 analysis: a professional could be liable for failing to use AI where a competent peer would have, and for using it without due diligence, testing or checking. What UK GPs have adopted and what they fear, why the duty to use travels with the duty to check, and the standard of care a firm can write before a court writes it.
  • Which professions face the greatest deskilling risk?
  • How will AI change medicine?
  • How will AI change law?
  • How will AI change consulting?
  • Will human judgement protect lawyers from AI?Not as a job description, and on that Richard Susskind is right. The hallucination record to date, Garicano's evidence that the work is leaving the firms that trained juniors, why handling uncertainty is not answering for it, a Crown Court judge's eight conditions before letting a tool into a trial, and what a firm can decide this quarter.
  • How will AI change accounting and audit?The oversight precedent everybody cites was replaced in April 2026, and its replacement puts generative AI expressly out of scope. Meanwhile the UK audit regulator found the six largest firms had not measured what their tools do to audit quality.
  • How will AI change journalism?Twenty-two broadcasters, 14 languages, 2,709 graded answers. Sourcing failed at 31 per cent against 20 per cent for accuracy, and the reputational cost arrives at the masthead that was cited rather than the assistant that cited it.
  • How will AI change customer service?The best-evidenced occupation there is, and the study watched the tool break. Gains ran to novices, the best agents got slightly worse, and the learning survived an outage only for the workers who had engaged with the suggestions.
  • Will AI replace teachers, and how will it change teaching?A school-randomised trial in England cut lesson planning time 31 per cent with no quality change a blinded panel could see. English teachers plan with it at 35 per cent and mark with it at 5 per cent, which is the profession quietly putting the tool where the risk is lowest.
  • How will AI change the public sector?Both flagship UK figures, 26 minutes a day and 19, are self-reported, and one was calculated from tick-box midpoints with its top tail capped by the analysts. The sector's deepest precedent is not a productivity study, it is the presumption that the computer is right.
  • How will AI change human resources?Retrieval models favoured White-associated names in 85.1 per cent of comparisons. The first algorithmic hiring audit law in the world produced published reports from 18 of 391 employers checked. And the function advising on capability is the one most exposed.
  • What happens to medical and legal training?Two professions that wrote down how many supervised repetitions qualification takes, meeting a technology that performs the repetitions. The three failures Nature Medicine separates, the colonoscopy result that moved thirty-year careers in months, the trial where a model sixteen points better than the doctors improved them by two, and the paragraph where the Divisional Court reached for the pupil barrister as its analogy for a machine.
  • Pricing Work When Production Is Cheap

Everyday life

International

  • AI and work, country by country
  • AI and work in AsiaSingapore has written the loss of entry-level training grounds into a national AI framework. Hong Kong measures augmented reality adoption and does not ask about AI at all. Japan, Korea and the adoption gap, all read at source.
  • AI and work in the GulfSaudi Arabia, the UAE and Qatar read at the issuing body's own pages. The one Gulf instrument that names over-reliance, the region's only official AI adoption statistic, and the widely quoted claims that could not be verified.
  • AI and work in JapanThe country with the strongest imaginable economic case for adopting AI, an 11 million worker shortfall by 2040, and adoption running at roughly 18 per cent. The control condition this debate never had.

Definitions

Reference and record

Where Rahim speaks on this

Rahim Hirji gives this argument as a keynote for boards and leadership teams in London, Europe and North America.

Reference · SS-2026-225

Cite this page

Hirji, R. (2026). The full index. The SuperSkills evidence base, SS-2026-225. https://thesuperskills.com/research/all.

An evidence review by Rahim Hirji, not peer-reviewed research. For a material claim, cite the underlying study as well; every study here carries its own permanent link.

How citations and IDs work
Ask the evidence
What does the evidence actually show?What should our board be asking about this?Where does Rahim disagree with the consensus?
Bring this into your organisation

If this describes something happening in your teams, say so.

Keynotes, board sessions and advisory work, drawing on research across more than 200 organisations in 30 countries. Tell me the room, the date and the shift you need. A reply within 24 hours.

Start a conversation

Topics and audiences  ·  All research

Box of Amazing

Rahim’s free weekly letter on AI and human capability

If this was useful, the weekly letter is where the thinking happens first. Most of what ends up on this site starts there. Every week, an essay and ten links to broaden your view of the world. Read by 25,000 people, every week since 2017. Free, and one click to stop.

Opens Substack to confirm. No pitch in it, unsubscribe in one click, and nobody follows up because you read something.

Running an event, or responsible for how AI arrives in your organisation? Keynotes  ·  Advisory  ·  Boards  ·  Enquire