Gen Z Knows AI Best and Trusts It Least: Scepticism Earned
There is a particular kind of person who opens a large language model a dozen times a day, leans on it to draft the awkward email and summarise the reading and untangle the spreadsheet, and who, if you asked them whether they trusted the thing, would look at you as though you had asked whether they trusted a vending machine. They use it. They do not believe in it. The two facts sit together without friction, because for this person they were never in tension to begin with.
That person is, statistically, most likely to be young. According to a Gallup survey of Americans aged 14 to 29, conducted between 24 February and 4 March 2026 and released that April, fifty-one per cent of Generation Z now uses generative artificial intelligence at least weekly. This is the cohort born between 1997 and 2012, the demographic that has folded these tools into study, work and daily life more thoroughly than any other. And yet, in the same survey, sixty-nine per cent of Gen Z workers said they placed more trust in work completed without AI, against twenty-eight per cent who placed it in work produced with AI's assistance. Only three per cent reserved their greatest trust for output generated solely by a machine. The people who use the technology most are the people who believe in it least.
This is not the story the industry told itself it was writing. The implicit promise of consumer AI was that familiarity would breed confidence, that once people saw what the systems could do the scepticism would melt into dependence and dependence into faith. The opposite has happened. Exposure has curdled into wariness. And the wariness is sharpest precisely where exposure is deepest. A Fortune analysis published in April 2026, drawing on research by the enterprise AI firm Writer and the consultancy Workplace Intelligence, found that forty-four per cent of Gen Z workers admitted to actively sabotaging their employer's AI rollout, against twenty-nine per cent of workers overall. These are not refuseniks standing outside the technology and lobbing stones. They are inside it, using it daily, and quietly undermining it at the same time.
The question this poses is not the tired one about whether AI is good or bad. It is stranger and more revealing. What does it tell us about the state of AI deployment in 2026 that the most technologically fluent generation alive has become its most committed sceptics? And if their doubt is not ignorance, but something closer to experience, what would it actually take for trust to be earned rather than simply assumed?
The Numbers Behind a Generational Rupture
It is worth being precise about the data, because the precision is where the story lives. The headline figures are frequently muddled in the retelling, and the muddle flattens something important.
The 2026 Gallup study, based on a probability sample of 1,572 young Americans with a margin of error of plus or minus 3.6 percentage points, did not merely record low trust. It recorded a slide. Among employed Gen Z respondents, forty-eight per cent said the risks of AI in the workforce now outweighed the benefits, up from thirty-seven per cent a year earlier, while only fifteen per cent judged the benefits greater. Excitement about the technology fell fourteen points in twelve months, to twenty-two per cent. Hopefulness dropped nine points. Anger rose nine, to thirty-one per cent. Curiosity remained the emotion this generation reports most often, at forty-nine per cent — but anxiety ran a close second at forty-two per cent, unmoved from the year before, holding its ground while the positive feelings collapsed around it. That is the shape of the shift, and it is more telling than a simple slump would be: a generation still interested, no longer optimistic. The same study found forty-two per cent judged AI harmful to critical thinking, against twenty-five per cent who found it helpful, and eighty per cent reckoned their own use of it likely to impair future learning. Perhaps most damningly, a separate survey published that May by the software firm GoTo and the consultancy Workplace Intelligence found that forty-six per cent of Gen Z workers said AI was making them “dumber,” compared with thirty-nine per cent of workers overall.
The broader adult population is not exactly a fountain of confidence either, though its scepticism is quieter. A YouGov poll of American adults conducted in early December 2025 found that just five per cent said they trusted AI “a lot.” Over the preceding year, twenty-five per cent reported that their trust in the technology had fallen, against twenty-one per cent who said it had risen — a net erosion at exactly the moment the tools were becoming inescapable. A separate Quinnipiac University poll published in late March 2026 found that seventy-six per cent of Americans trusted AI only rarely or sometimes, that seventy per cent believed it would reduce job opportunities, and that seventy-six per cent felt businesses were not being transparent about how they used it. The share of employed Americans who worried their own job would become obsolete had risen to thirty per cent from twenty-one per cent a year before.
Set these numbers beside the adoption curves and the dissonance becomes almost vertiginous. Just under half of American adults now use AI chatbots — forty-nine per cent, up from a third in 2024. Usage is climbing steeply. Trust is not merely failing to climb with it. Trust is going the other way. The two lines have decoupled, and the wider the gap grows the more it demands explanation, because in almost every historical case familiarity with a technology has tracked comfort with it. Cars, aeroplanes, the internet, the smartphone — each followed a rough arc in which early fear gave way to routine as competence accumulated. AI, so far, is refusing to trace that arc. It is being adopted and distrusted at the same time, by the same people, with the intensity of both rising in parallel.
Why Adoption Was Never the Same as Endorsement
The industry's confusion on this point stems from a category error that runs deep in how technology companies read their own metrics. They treat usage as a verdict. If the numbers go up, the product is winning; the market has spoken; the doubters will come around. This is a serviceable heuristic for a video game or a food-delivery app, products people choose freely and abandon freely, where continued use really is a decent proxy for satisfaction.
It is a terrible heuristic for AI, because AI is increasingly not chosen. It is arriving pre-installed. It sits at the top of the search results whether or not you asked for a generated summary. It is stitched into the word processor, the email client, the customer-relationship platform, the operating system. It is mandated by the employer who has bought an enterprise licence and set adoption targets that show up in performance reviews. When a tool is embedded in the infrastructure of daily work and daily life, using it stops being an endorsement and becomes something more like breathing the available air. You do it because it is there, not because you have appraised it and found it worthy.
This is the phenomenon that the data on Gen Z brings into focus with unusual clarity. The Brookings Institution analysis by Josie Stewart and Brooke Tanner, published on 22 July 2026 under the pointed title “Policy—not PR—will determine Gen Z's trust in AI,” makes the underlying dynamic explicit. Stewart, a senior research and communications assistant, and Tanner, a research analyst, note that around half of Gen Z uses generative AI at least weekly while expressing some of the deepest reservations of any group — a pattern they read not as confusion but as coherence. High use and low trust are not, in their account, a contradiction to be resolved. They are the natural product of a situation in which people are compelled into contact with a technology they have had ample opportunity to appraise, and have appraised without illusion.
The distinction between adoption and endorsement matters because the industry keeps mistaking the first for the second and then acting surprised when the sabotage figures come in. Forty-four per cent of Gen Z workers undermining an AI rollout is not the behaviour of a market that has spoken in favour. It is the behaviour of people who have been given a tool they did not ask for, told to use it, and found the quiet forms of resistance — feeding it junk, routing around it, declining to trust its output — that remain available when the loud form, refusal, has been taken off the table. Adoption, in this light, is not the end of the trust problem. It is the container in which the trust problem is now being fought.
A Distrust That Was Learned, Not Inherited
The comforting explanation for all this, and the one the technology sector has reached for most often, is that the scepticism is a deficit — of knowledge, of literacy, of exposure. Young people are anxious, the reasoning goes, because they do not understand the technology well enough to see its promise, and the remedy is education: more AI-literacy programmes, better onboarding, clearer communication about capabilities and limits. It is a flattering theory for those who hold it, because it locates the problem entirely in the user and requires nothing of the product.
The theory has one fatal weakness. The people expressing the deepest doubt are the ones who understand the technology best. A September 2025 Pew Research Center survey found that sixty-two per cent of adults under thirty had heard “a lot” about AI, the highest of any age group — and this same, best-informed cohort was the most pessimistic, with sixty-one per cent saying the technology would worsen people's ability to think creatively and fifty-eight per cent saying it would erode their capacity to form meaningful relationships, both figures around twenty points higher than among those over sixty-five. By February 2026, a further Pew survey had stated the whole decoupling in one cohort's own numbers: sixty-six per cent of adults aged 18 to 29 used AI chatbots and thirty-one per cent used them daily, making them the heaviest-using age group in the country, while just fourteen per cent expected AI's effect on society to be positive and forty-eight per cent expected it to be negative. The heaviest users are the least persuaded, and by some distance. The scepticism does not recede as understanding grows. It deepens. This is the single most inconvenient fact for the literacy theory, and it is decisive: you cannot educate people out of a conclusion they reached by paying attention.
The Brookings analysis and a companion Fortune essay by the outlet's editor Nick Lichtenberg, published on 16 April 2026, converge on the same reframing. The distrust is not irrational. It is earned. It reflects first-hand experience of a specific and repeated kind. Gen Z has watched AI systems state falsehoods with total confidence, the phenomenon the field politely calls hallucination and everyone else calls being wrong without knowing it. They have generated content with these tools and then been penalised for it by institutions — more than half of college students report that their schools either discourage AI use, at forty-two per cent, or ban it outright, at eleven per cent, even as sixty-three per cent of faculty concede that their 2025 graduates were not prepared to use AI in the workplaces that now demand it. They have been told the technology is a co-pilot and then watched it fold the entry-level jobs where they expected to build the very judgement that would let them supervise it.
There is a body of research that lends this lived experience empirical teeth. A June 2025 preprint from the MIT Media Lab reported weaker neural connectivity and poorer recall among participants who wrote essays with the help of a large language model, compared with those who wrote unaided. A study by Microsoft Research and Carnegie Mellon University, surveying 319 knowledge workers across 936 AI-assisted tasks, found that higher confidence in a generative system was associated with less critical thinking, not more. Work by Michael Gerlich at the SBS Swiss Business School in 2025 found that heavier AI use correlated with greater cognitive offloading and lower critical-thinking scores, an effect most pronounced among younger participants. None of this proves that AI makes people less capable in any simple causal sense. But it means that when nearly half of Gen Z reports the technology is making them “dumber,” they are describing something that researchers are independently measuring. The feeling has a footing.
What emerges is a distrust that was not inherited from anxious parents or absorbed from alarmist media, but assembled, piece by piece, from direct encounter. And it is active rather than passive. A Skyword survey of a thousand American consumers, published in June 2026, found that sixty-seven per cent of Gen Z respondents were using AI more than they had a year earlier, against fifty-two per cent of consumers overall, and that nearly one in three had contacted a brand directly to correct something an AI tool had said about it — close to double the rate of the general population. They do not merely doubt the output. They check it, and then they go to the trouble of reporting what they find. It is the scepticism of the mechanic who has looked under the bonnet, not the passenger who is nervous about the noise.
The Quiet Erasure of the First Rung
If there is a single grievance that anchors the generational rupture, it is the disappearance of the entry-level job, and it deserves to be understood on its own terms, because it connects the abstract question of trust to something concrete and material.
The traditional path into skilled work runs through a period of supervised incompetence. The junior analyst builds the model badly and is corrected. The trainee lawyer drafts the memo, has it torn apart, and drafts it again. The apprentice does the tedious, low-stakes work under the eye of someone who has done it a thousand times, and in the doing acquires the judgement that eventually lets them do the high-stakes work and, later still, supervise the next apprentice. The tedium is not incidental to expertise. It is the mechanism by which expertise is transmitted. Remove the bottom rung and you do not simply inconvenience the people standing on it. You break the ladder.
This is precisely the function that generative AI is best at absorbing. The summarising, the first drafts, the routine research, the boilerplate code — the substance of entry-level work is also the substance of what these systems do most competently. The consequence is measurable. Lichtenberg's Fortune analysis reports that junior hiring fell nearly eight per cent within six quarters at companies adopting AI, and that unemployment among recent college graduates reached 5.7 per cent in the fourth quarter of 2025, exceeding the national rate, with underemployment among recent graduates standing at 42.5 per cent, the highest since 2020. The displacement is arriving not through dramatic mass layoffs but through what Lichtenberg calls “quiet erasure” — the junior role that is never posted, the pipeline that thins one unfilled vacancy at a time.
The concern here is old, even if the technology is new. In 1974 the Marxist theorist Harry Braverman, in his study of the twentieth-century workplace, described the way industrial management systematically stripped skill out of jobs, concentrating judgement in a shrinking managerial layer while the majority were left with degraded, deskilled tasks. In 1983 the ergonomics researcher Lisanne Bainbridge, in a paper on automation that has only grown more relevant, identified what she called the “ironies of automation”: that the more you automate a system, the more critical and the more difficult the residual human role becomes, because the human is now expected to monitor a process they no longer routinely perform and to take over precisely in the moments the machine cannot handle — the moments that require exactly the fluent competence the automation has prevented them from maintaining. A generation that has been handed AI to do its formative work, and then told it will one day be expected to oversee that AI, has read Bainbridge without needing to read Bainbridge. It has intuited the trap. You cannot supervise what you were never allowed to learn.
This is why the deskilling anxiety is not nostalgic or self-pitying. It is a structurally sound observation about how competence is built and how it is being interrupted. When a young worker distrusts a system that produces confident output while quietly eroding the conditions under which they might learn to check that output, they are not being irrational. They are noticing the mechanism.
When Opting Out Stops Being an Option
The most under-examined aspect of this whole predicament is the one the brief that prompted this essay named directly: millions of people, disproportionately young, now use AI tools daily not because they trust them but because opting out is no longer practically available. The condition deserves a name of its own — compelled adoption — because it inverts the usual relationship between a technology and its user.
Consider the ordinary surfaces of a working life in 2026. The search engine returns an AI-generated summary above the links, so that even the act of looking something up now routes through a generative system whether or not you wanted it to. The office suite offers to draft, rewrite and summarise, and increasingly assumes you will accept. The employer has purchased seats and set targets; the Bentley-Gallup research on business AI has long shown that adoption is being driven from the top down, and the Writer and Workplace Intelligence survey underlying the sabotage figures polled 2,400 knowledge workers precisely because their firms were rolling the tools out to them, including 1,200 executives doing the rolling. To decline, in this environment, is not a neutral act of consumer preference. It is to fall behind colleagues, to miss targets, to mark yourself as a resister in an organisation that has decided the future is settled.
Compelled adoption changes the meaning of every usage statistic the industry cites. When a company boasts that a tool has reached fifty per cent weekly penetration among young workers, it is describing, in part, the success of its own mandate, not the freely given approval of its users. And it changes the meaning of resistance, too. If you cannot refuse the tool, the only forms of dissent left are the small, deniable ones the sabotage research catalogues: feeding proprietary information into public models where it should not go, using unapproved tools instead of the sanctioned one, quietly producing lower-quality AI output to make the system look less effective, tampering with the metrics that would show it succeeding. These are not the tactics of people who have been persuaded. They are the tactics of people who have been conscripted and are looking for the exits.
There is something almost poignant in the figure of the reluctant user — the person who has internalised the futility of refusal so completely that they no longer even frame it as a choice, who reaches for the tool with one hand while withholding belief with the other. This is not the enthusiastic early adopter of technology mythology, nor the noble refusenik of the resistance narrative. It is a third thing, less legible and more common: the person who complies and does not consent, who has separated the practical question of use from the moral question of trust because the practical question was decided for them. Millions of people now live in this posture, and the industry's dashboards cannot see them, because the dashboards were built to count clicks, not to detect the quiet withdrawal of faith behind them.
The Long History of Trusting Machines
None of this is quite as unprecedented as the breathlessness of the moment suggests, and the historical parallels are worth drawing not to diminish the present but to calibrate it.
The Luddites of 1811 to 1816 have been so thoroughly reduced to a slur — a Luddite is now simply a fool who fears progress — that their actual grievance has been buried. The framework-knitters and croppers who broke machines in the English Midlands were not against technology as such. Many were skilled operators of the machinery of their trade. Their objection was to a specific deployment of new machines by owners who used them to drive down wages, circumvent labour standards and concentrate the gains of higher productivity in their own hands, while the workers who had built the industry's skill base were cast off. Their quarrel, in other words, was not with the loom but with the distribution of the loom's benefits and the absence of any say over how it was introduced. Read the contemporary data on AI — the seventy per cent who expect it to cut jobs, the sabotage, the demand for transparency and recourse — and the rhyme is unmistakable. The distrust has never really been of the machine. It has been of the arrangement around the machine.
The twentieth century added a subtler lesson. As automation spread through cockpits and control rooms, researchers documented a phenomenon they called automation bias: the human tendency to over-trust an automated system, to defer to its output even when it was wrong and even when contrary evidence was available. The danger of automation, it turned out, was rarely that people rejected it. It was that they surrendered to it too completely, ceding judgement to a machine that did not deserve the deference. Seen against this history, the Gen Z posture — high use, low trust, judgement withheld — starts to look less like a pathology and more like a hard-won corrective. A generation that refuses to grant automation the deference it has not earned is a generation that has, perhaps unknowingly, absorbed the central safety lesson of the automation age.
And then there is the economist Robert Solow's famous observation from 1987, at the height of the office-computing boom, that “you can see the computer age everywhere but in the productivity statistics.” The productivity paradox he named has an eerie contemporary echo. For all the money and disruption, the aggregate productivity gains from generative AI remain, as of 2026, stubbornly hard to locate in the macroeconomic data. Transformative technologies, from the steam engine to electrification to the personal computer, have historically taken decades to deliver their promised gains, precisely because realising them requires reorganising the institutions, skills and trust relationships around the machine — the very things that are currently in open dispute. The trust deficit is not a delay before the revolution. It may be part of the mechanism by which any real gains are either eventually secured or permanently squandered.
What Recourse Would Actually Look Like
If the distrust is earned, then the standard remedies are worse than useless, and this is the sharpest edge of the Brookings argument. Stewart and Tanner are explicit that the trust deficit will not be closed by better marketing or another round of AI-literacy campaigns, because the deficit is not a communications failure. You cannot advertise your way out of a conclusion that people reached through experience. What would close it, they argue, is substantive change that gives people meaningful oversight and genuine recourse over the systems that affect their lives — the ability not merely to be told an AI was used, but to contest its output, to reach a human, to seek redress when it errs.
The tech industry's own gestures, Stewart and Tanner note, have been telling in their inadequacy: chief executives pledging to cover the electricity costs of their data centres, walking back their more lurid layoff predictions, offering voluntary reassurances. These are the moves of institutions that understand there is a problem and hope it can be managed with goodwill rather than obligation. But the same research documents why goodwill will not suffice. Over the past decade, Americans' confidence in technology firms has fallen faster than their confidence in most other institutions, and the youngest cohort has lost faith in the major tech companies faster than any other group. These are the same firms now asking to be trusted. Voluntary commitments from actors whose credibility is itself in freefall are not a foundation on which trust can be rebuilt.
The alternative is binding. Some of the scaffolding already exists. The European Union's AI Act, which entered into force in 2024 and is being phased in across the middle of the decade, takes a risk-based approach, banning certain uses outright, imposing strict obligations on high-risk systems, and requiring, among other things, transparency about when people are interacting with a machine. The Union's older data-protection regime, the GDPR, already grants individuals a right, under its provisions on automated decision-making, not to be subject to purely automated decisions that produce legal or similarly significant effects, together with a right to obtain human intervention and to contest the outcome. These are, in embryo, exactly the recourse mechanisms the trust deficit demands: not a promise that the system is fair, but an enforceable route to challenge it when it is not.
The Brookings analysis sketches where such obligations would bite hardest, across four domains. On labour, it points to portable benefits, wage insurance for displaced workers, lifelong-learning accounts, honest disclosure standards around AI-attributed layoffs, and targeted support for the entry-level cohort bearing the early costs of the transition. On creativity, a framework for consent and compensation, so that the people whose work trained these systems are not simply expropriated. On the environment, mandatory energy and water disclosures for data centres and protections so that households do not quietly subsidise private compute through higher utility bills. On safety, privacy- and safety-by-design requirements, anti-discrimination enforcement and heightened protections for minors. The common thread is not hostility to the technology. It is the insistence that the people affected by a system should have some enforceable purchase on it — a lever, a court, a human being who can be held to account.
This is the substance of what “earned” would mean. Trust, in any durable sense, has never been a feeling that can be induced by persuasion. It is a wager about the future behaviour of another party, and it is rational only to the extent that the other party can be held to its word. We trust the surgeon because the surgeon is licensed, regulated, insured and liable. We trust the aeroplane because the aviation system is saturated with accountability, from the certification of the airframe to the investigation of every incident. We do not trust these systems because we were told they were trustworthy. We trust them because recourse exists when they fail, and because that recourse has teeth. AI, for now, has been asking for the trust without building the accountability. The generation that has looked most closely has noticed the asymmetry, and declined.
The Verdict of the People Who Know It Best
Return, at the end, to the reluctant user — the young worker with the model open and the belief withheld. The instinct of the industry, and of a good deal of the commentary around it, has been to treat this person as a problem to be solved: too anxious, too cynical, insufficiently visionary, in need of reassurance or re-education. This essay has argued for the opposite reading. The reluctant user is not a problem. They are a signal, and possibly the most reliable one available.
Because they are the people who know the technology best. They have used it more, understood it more, encountered its failures more directly than any focus group or executive or optimistic keynote. When the best-informed users of a tool converge, in growing numbers and with rising intensity, on the judgement that it is useful but not to be trusted, that its output must be checked and its incursions resisted, they are not exhibiting a deficit. They are rendering a verdict. And the verdict is not that the technology is worthless — they would not use it daily if it were — but that it has not yet earned the deference it keeps demanding.
That verdict is, in a sense, good news, though it will not feel like it to the companies whose valuations depend on the deference. A generation that uses a powerful technology without surrendering its judgement to it is a generation behaving exactly as a healthy society ought to behave in the presence of something new and consequential. The scepticism is not the failure of AI adoption. It is the immune response of people who have seen enough to know that adoption and trust are different things, and that the second must be earned on terms the first cannot dictate.
The uncomfortable implication for the industry is that there is no shortcut. The trust cannot be marketed into existence, cannot be assumed, cannot be extracted by mandate from users who have already appraised the goods. It can only be built the slow way, through systems that give people real oversight and real recourse, through institutions that make the machine's makers answerable when the machine fails, through a distribution of the technology's benefits that the people generating those benefits can recognise as fair. Until then, the reluctant users will keep doing what they are doing: using the tool, withholding the faith, and waiting — reasonably, patiently, with their judgement intact — to be given a reason to believe.
References
Gallup (2026) “Gen Z's AI Adoption Steady, but Skepticism Climbs.” Web survey of 1,572 Americans aged 14–29, conducted 24 February – 4 March 2026, margin of error ±3.6 percentage points. Available at: https://news.gallup.com/poll/708224/gen-adoption-steady-skepticism-climbs.aspx
GoTo and Workplace Intelligence (2026) “The Pulse of Work in 2026: Opportunity, Risk, and Responsibility in an AI-Driven Workplace.” Survey of 2,500 global employees and IT leaders, released 19 May 2026. Available at: https://www.goto.com/blog/pulse-of-work-2026
Stewart, J. and Tanner, B. (2026) “Policy—not PR—will determine Gen Z's trust in AI,” Brookings Institution, 22 July 2026. Available at: https://www.brookings.edu/articles/policy-not-pr-will-determine-gen-zs-trust-in-ai/
Angelo, J. (2026) “Gen Z workers who fear AI will take their job are actively sabotaging their company's AI rollout,” Fortune, 8 April 2026. Survey by Writer and Workplace Intelligence of 2,400 knowledge workers (including 1,200 C-suite executives) across the US, UK and Europe. Available at: https://fortune.com/2026/04/08/gen-z-workers-sabotage-ai-rollout-backlash/
Lichtenberg, N. (2026) “Gen Z turning its back on AI isn't irrational—it's a verdict on everyone who failed them,” Fortune, 16 April 2026. Available at: https://fortune.com/2026/04/16/why-does-gen-z-distrust-ai-anxiety-failures-college-work-government/
YouGov (2025) “Most Americans use AI but still don't trust it.” Surveys of US adults conducted 4 and 8 December 2025 (n=1,287 and n=1,187). Available at: https://yougov.com/en-us/articles/53701-most-americans-use-ai-but-still-dont-trust-it
Quinnipiac University Poll (2026), reported in TechCrunch, “As more Americans adopt AI tools, fewer say they can trust the results,” 30 March 2026. Available at: https://techcrunch.com/2026/03/30/ai-trust-adoption-poll-more-americans-adopt-tools-fewer-say-they-can-trust-the-results/
Pew Research Center (2025) “How Americans View AI and Its Impact on People and Society,” 17 September 2025. Fielded 9–15 June 2025, n=5,023. Available at: https://www.pewresearch.org/science/2025/09/17/how-americans-view-ai-and-its-impact-on-people-and-society/
Pew Research Center (2026) “Americans and AI 2026: Chatbots, Smart Devices and Views on Impact,” 17 June 2026. Fielded 17–23 February 2026, n=5,119. Available at: https://www.pewresearch.org/internet/2026/06/17/americans-and-ai-2026-chatbots-smart-devices-and-views-on-impact/
Pew Research Center (2026) “How Americans' opinions and use of AI differ by age,” 17 June 2026. Available at: https://www.pewresearch.org/internet/2026/06/17/how-opinions-and-use-of-ai-differ-by-age/
Kosmyna, N. et al. (2025) “Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task,” MIT Media Lab preprint, June 2025. Available at: https://www.media.mit.edu/publications/your-brain-on-chatgpt/
Lee, H-P. et al. (2025) “The Impact of Generative AI on Critical Thinking,” Microsoft Research and Carnegie Mellon University. Survey of 319 knowledge workers across 936 AI-assisted tasks.
Gerlich, M. (2025) “AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking,” SBS Swiss Business School, published in Societies.
Skyword (2026) “AI and Brand Trust survey.” Survey of 1,000 US consumers, published 29 June 2026. Reported at: https://www.contentgrip.com/gen-z-ai-brand-trust-skyword-survey/
Braverman, H. (1974) Labor and Monopoly Capital: The Degradation of Work in the Twentieth Century. New York: Monthly Review Press.
Bainbridge, L. (1983) “Ironies of Automation,” Automatica, Vol. 19, No. 6, pp. 775–779.
Solow, R. (1987) “We'd Better Watch Out,” New York Times Book Review, 12 July 1987.
Bentley University and Gallup (2023–) “Business in Society: The Bentley-Gallup Report on AI.” Available at: https://www.bentley.edu/gallup/ai
European Union (2024) “Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act).” Available at: https://eur-lex.europa.eu/eli/reg/2024/1689/oj
European Union (2016) “Regulation (EU) 2016/679 (General Data Protection Regulation),” Article 22 on automated individual decision-making. Available at: https://eur-lex.europa.eu/eli/reg/2016/679/oj
Tim Green UK-based Systems Theorist & Independent Technology Writer
Tim explores the intersections of artificial intelligence, decentralised cognition, and posthuman ethics. His work, published at smarterarticles.co.uk, challenges dominant narratives of technological progress while proposing interdisciplinary frameworks for collective intelligence and digital stewardship.
His writing has been featured on Ground News and shared by independent researchers across both academic and technological communities.
ORCID: 0009-0002-0156-9795 Email: tim@smarterarticles.co.uk
Listen to the free weekly SmarterArticles Podcast