Why young Americans are turning against AI

Young people are not happy about AI. At American graduation ceremonies in May, University of Arizona graduates repeatedly booed former Google chief executive Eric Schmidt as he described AI’s promise. Students at the University of Central Florida jeered when a speaker called AI “the next industrial revolution”. Days later, in contrast, comedian Ronny Chieng received cheers at Harvard University after telling graduates that their generation’s mission was to “destroy AI”.

The reaction extends beyond commencement crowds. In a 2026 Gallup survey of 1,572 Americans aged 14 to 29, excitement had fallen from 36% to 22% in one year and anger had risen from 22% to 31%. Young people are entering adulthood as institutions reorganise work around tools they did not ask for. They reasonably wonder whether AI will help them build careers or make it impossible to build skills and find meaningful work.

Why the backlash against AI?

Young people object to AI on several fronts. They worry about the energy and water consumed by data centres. Writers, designers and musicians see systems trained on human work now competing with its creators.

They also encounter “AI slop”: cheap, generic content flooding feeds, search results and professional platforms. One LinkedIn user described the site as “absolutely riddled with AI slop”. The feeling that now everyone sounds the same is supported by research led by Kibum Moon. We analysed 372,793 college-admissions essays. After ChatGPT’s release, essays became more varied in their word choices but their underlying ideas became more homogeneous.

When it comes to young people beginning their careers, there are three concerns that trump all others.

The first is the worry that AI will erode skills and critical thinking. In a nationally representative survey my colleagues and I conducted among American Gen Z, 79% said AI makes people lazier and 62% said it makes people less smart.

Research supports two paths to these outcomes. First, AI can remove the practice needed for skills to develop. Across three randomised trials involving 1,222 people, AI improved performance while it was available, but users performed worse and gave up sooner after it was removed. Second, fluent answers can discourage scrutiny. In three experiments on “cognitive surrender”, access to accurate AI raised reasoning accuracy by 25 percentage points. But when the AI was faulty, people simply deferred to the AI, resulting in a performance drop of 15 percentage points. Participants gained confidence even when AI led them astray.

The second is the worry that AI erodes agency: the feeling that we direct our actions and can shape what happens. Agency makes work feel like ours and supports motivation and meaning. In a preregistered experiment my colleagues and I ran with 1,533 Americans, participants wrote cartoon captions with or without AI. AI produced a very large drop in their sense of control over the creative process, even as it increased their sense of achieving the desired outcome. Other researchers similarly found across four experiments involving 3,562 people that AI improved immediate performance but could undermine subsequent intrinsic motivation. Better results can arrive with a weaker sense of the meaningfulness of work.

The third is the worry that AI will make it impossible for young people to enter the labour market. The clearest evidence points to a change in who gets hired rather than mass unemployment. Researchers studying Sweden’s entire workforce and 4.6 million job advertisements found little evidence that ChatGPT caused the broad decline in vacancies; that decline began when interest rates rose. Yet within the same employers, employment among 22-to-25-year-olds in the most AI-exposed occupations fell 5.5% relative to less-exposed roles by early 2025. The adjustment came overwhelmingly through reduced hiring, while workers aged 31 to 49 were essentially unaffected. Recent American research finds a similar pattern. AI may therefore close doors to newcomers long before it removes incumbent workers.

Today’s savings, tomorrow’s shortage?

When companies replace entry level jobs with AI, it creates the apprenticeship problem. Companies can use AI to absorb the research, drafting and coding once assigned to junior employees. Those assignments taught newcomers the skills needed to ascend the organisational chart. They also built relationships with senior colleagues who supplied feedback and eventually entrusted juniors with harder decisions. If firms remove the bottom rung of the ladder, they weaken the pipeline of people prepared to climb it. Today’s savings can become tomorrow’s shortage of experienced workers.

Gen Z’s dislike for AI makes sense: AI can weaken skills developed through work, reduce the meaning derived from doing it and narrow access to the jobs that lead to professional growth. These concerns are especially poignant to them, compared to their older counterparts who already have the skills, might get meaning from mentoring or other tasks that AI can’t easily replace, and who have already climbed the organisational ladder.

Despite all this, workers and organisations face a trade-off. Delegating work to AI might make users more productive in the short term, but in the long term, they may see their skills and their motivation erode. Organisations preferring AI agents over entry level workers might see productivity gains and cost savings in the short term, but might have a thinner talent pipeline years later.

This trade-off exploits a basic cognitive limitation. Humans show present bias: we overweight immediate rewards and discount delayed consequences. Each act of delegation offers a visible and certain payoff — finishing this email, analysis or piece of code sooner. The possible cost is cumulative, uncertain and borne by a future self or a future employer.

Workers should therefore be intentional about what to delegate. They should hand off tasks that are unrelated to the capabilities they value. For skills that matter, either limit delegation or use AI in a way that preserves learning, even if that takes longer. A junior analyst might let AI format a chart but interpret it independently. A programmer might draft a solution before asking AI to review it.

Use AI, maintain critical thinking

Staying in the loop and iterating with AI rather than asking for finished outputs is also a way to preserve learning and meaning. Ask it for examples, criticism, explanations or a sceptical counterargument. In our experiments on professional writing, people who practised with AI exerted less effort yet later wrote better cover letters without it. A follow-up experiment showed why: a strong AI revision provided an example users could learn from. 

Employers must make the same trade-off consciously. If AI completes junior tasks, firms should replace the learning those tasks supplied: require first-pass reasoning before AI assistance, pair new workers with experienced reviewers, rotate them through consequential decisions and assess whether they can perform core tasks independently. Productivity metrics shouldn’t just myopically track how many minutes are shaved off of a process, but also how much human capital is being built (or eroded).

Present bias aside, choosing which capabilities to protect is difficult for yet another reason. Workers cannot know which capabilities AI will make obsolete and which will become more valuable precisely because AI is ubiquitous. Long division became dispensable after calculators; numerical judgment did not. Likewise, routine drafting or coding may disappear as tasks while still providing the practice that builds taste, error detection and professional judgment. A young worker deciding what to delegate is therefore betting on a labour market that does not yet exist.

Gen Z is already accepting and using AI. Its distrust is a demand for a better future of work — one in which efficiency today preserves expertise, agency and opportunity tomorrow. Employers and technology companies should treat that demand as useful feedback while there is still time to adapt.

Related: [Big read] China’s young workers pay the price of AI before reaping the gains | How AI is rewiring white-collar work in China