The booing kicked off before Gloria Caulfield could finish her sentence. At the University of Central Florida’s College of Arts and Humanities graduation on May 8th, real estate exec Caulfield told the graduating class that “the rise of artificial intelligence is the next industrial revolution.” One student yelled, “AI sucks”, and the crowd erupted. Caulfield turned to the other speakers onstage, visibly confused: “OK, I struck a chord. May I finish?”
This wasn’t a one-off. At Middle Tennessee State University, Big Machine Records CEO Scott Borchetta was booed for telling graduates to “deal with it” because “it’s a tool”. At the University of Arizona on May 18th, former Google CEO Eric Schmidt was hissed at the moment he mentioned AI, prompting him to make a statement that has been read as either ironic or deeply un self-aware: “There is a fear in your generation that you are inheriting a mess that you did not create, and I understand that fear.”
And at Glendale Community College in Arizona, the institution itself deployed an AI system to read graduates’ names during the ceremony. It skipped dozens, mispronounced others, and displayed the wrong names on screen. The college president told the crowd it was “a lesson learned”, and the crowd booed again. Graduating student Grace Reimer pointed out the irony: her class syllabi contained strict rules about AI, including the possibility of expulsion for using it, yet the college had used it on her graduation day.
The 2026 US graduation season has become an unplanned referendum on how young people feel about GenAI, and the verdict is loud, public, and probably pretty uncomfortable for anyone building an institutional AI strategy.
But the same week Eric Schmidt was being booed in Arizona, the Lumina Foundation-Gallup 2026 State of Higher Education study reported that 57% of US college students use AI tools in their coursework weekly and 20% use them daily. Princeton University faculty voted to scrap their 133-year-old honour code and proctor all in-person exams after a senior survey found 29.9% of students admitted to cheating and only 0.4% had ever reported a peer. Stanford senior Theo Baker, writing in the New York Times, described cheating as “omnipresent” and said he didn’t know a single person who hadn’t used AI to get through an assignment.
Learning First’s May 2026 report, AI Use in Schools: Taking Action Now, reflects on over 3000 Australian teacher’s responses to the situation. The report suggests that “cognitive outsourcing” amongst students is rife, and that schools are struggling to keep up with the pace of change. The media in Australia has predictably jumped onto the report, posing AI as an “urgent threat” to senior assessment programs in K-12 and demanding that we “Get AI out of schools“.
The problem, of course, is much more nuanced than the headlines suggest. Dealing with the implications of AI in education is not simply a matter of banning technology in the classroom or returning to invigilated exams. And at the heart of the problem, the people most affected by AI in learning are telling us that they are deeply conflicted about the technology.
Students hate AI. And they cannot stop using it.
How are students using AI?
Despite the nascency of GenAI, the evidence on student use is now substantial, drawn from multiple large-scale surveys across different countries, and the patterns are remarkably consistent. Students are not, for the most part, just copy/pasting AI-generated essays into submission portals. They are doing something more varied, more pragmatic, and harder to categorise.
The headline numbers come from large self-reported surveys that need handling with obvious care. The HEPI/Kortext Student Generative AI Survey 2025 (of 1,041 UK undergraduates) found 92% of full-time undergraduates use GenAI and 88% have used it for assessed work, up from 66% and 53% the previous year, with the most common uses being explaining concepts, summarising academic material, and structuring ideas. Of the students surveyed 18% reported including AI-generated text directly in their submitted work.
The Digital Education Council’s Global AI Student Survey (3,839 students across 16 countries) found 86% using AI for academic tasks and 54% using it weekly, while Pew Research Center’s February 2026 data (1,458 US teenagers) showed 64% had used AI chatbots and 54% were using them specifically for schoolwork help, up from 13% in 2023. HEPI’s survey is run in partnership with an edtech vendor and the Digital Education Council is itself an industry membership organisation, so these particular headline percentages are best read as indicative of direction and rough magnitude rather than precise prevalence; they are also snapshots of a fast-moving situation, and the year-on-year jumps suggest the figures will be out of date almost as soon as they are published.
Increasingly though, academic literature is adding depth to the volume, and in many ways it corroborates these student surveys. Henderson, Bearman, Chung, Fawns, Buckingham Shum, Matthews and de Mello Heredia (2025), drawing on roughly 6,960 quantitative responses and 8,642 open-ended responses across Deakin, Monash, the University of Queensland and UTS, found that about half of students had sought feedback from GenAI. Among those who had, the vast majority found both AI feedback and teacher feedback helpful, but roughly 9 in 10 rated teacher feedback as trustworthy compared with only about 6 in 10 for AI.
Students valued GenAI for ease of access, timeliness, volume, and the perception that it was less risky than approaching a teacher, while distrusting it for reliability, contextual understanding, and disciplinary expertise; the authors concluded that AI and teacher feedback are “complementary but not interchangeable.” Other studies emerging from the same Australian project push further into the details. Bearman et al. (2025), working with 79 students in focus groups across four universities, found students consistently positioned their educational work as something that should come from and be owned by themselves, prioritising their own developing moral positions over institutional rules.
Oberg et al. (2026), reporting on a national survey of 8,021 students from the project, found scepticism was the most commonly identified emotion (55.8%) and that relief, guilt, gratitude, and vigilance circulated around AI in relation to assessment, learning, and creativity and voice. Chung et al. (2026), drawing on the same dataset, traced how students describe the usefulness of GenAI in pragmatic rather than enthusiastic terms.
These two kinds of evidence concur on the headline finding that use is high and rising, but they tell different stories about what that use means. The vendor-adjacent surveys show breadth and uptake; the Australian project shows that uptake coexists with moral discomfort, scepticism, and a consistent student preference for teacher relationships over machine outputs.
So, the picture is not one of mass laziness or wholesale fraud, despite what some headlines suggest. Students are using GenAI to explain things they do not understand, to get faster feedback than their teachers can provide, to structure their thinking before they write, and to manage workloads that feel unmanageable. Some are also using it to cheat, and the evidence on that is clear, growing, and impossible to ignore. But the majority use case is pragmatic and study-oriented, with students themselves often expressing discomfort about the very dependence the data documents.
Why can’t they stop?
Three pressures are converging on students simultaneously, and all of them push toward AI use even when students say they wish they didn’t have to.
Labour market pressure
Students are graduating into the tightest entry-level job market in years. The Washington Post reported in April 2026 that computer science enrolment at four-year US universities dropped 8.1% in 2025-26. A Zety/HR Dive survey of 1,000 Gen Z workers found nearly three in four said AI will reduce entry-level corporate job opportunities in the next five years, and 65% said a college degree would not protect them from AI-related job loss. Inside Higher Ed reported Gallup data showing 16% of currently enrolled college students had already changed their major because of AI, and 42% of bachelor’s degree students had reconsidered theirs.
Anecdotally, I have spoken to many Year 11 and 12 students here in Australia who feel the same pressures. Students who were considering pathways through “at risk” industries like graphic design and computer science are concerned about the future of the subjects and their job prospects. International students from Asia feel that if they don’t use AI in their Australian senior secondary years, they’ll return home and find themselves unable to compete with their peers.
Students are being told simultaneously that AI will take their jobs and that they need to learn AI to keep them; the rational response, however reluctant, is to use the technology.
Unclear boundaries
Direct quotes in the HEPI survey capture the frustration of students faced with confusing internal policies: “They dance around the subject. It’s not banned but not advised, it’s academic misconduct if you use it but lecturers tell us they use it. Very mixed messages.” The Digital Education Council survey found only 5% of students felt fully aware of comprehensive AI guidelines. When the rules are unclear, the penalties are uncertain, and the technologies are freely available, use becomes the default position for many students.
Again, my conversations with secondary school students back this up. Last year, I wrote about the two main reasons these students provided for their GenAI use: it’s good enough, and it’s better than me. The “good enough” students in particular were often skirting around vague or unclear school policies, treading the line between “acceptable use” and “what they could get away with”. Those conversations revealed bigger-picture issues including the pressure to complete in assessments, engagement (or lack thereof), and aspects of secondary education which transcend AI problems.
Peer Pressure
Baker’s Stanford essay described a campus where everyone knows everyone uses AI, nobody reports it, and the incentive structure rewards those who game the system. Princeton’s data showed the social enforcement/honour code mechanism had collapsed entirely: 44.6% of seniors had witnessed cheating but only 0.4% reported it. In an environment where non-use feels like unilateral disarmament, even students who are uncomfortable with AI feel they cannot afford to opt out. The Gallup/Walton survey found students whose schools explicitly allow AI are 25 percentage points more likely to feel prepared to use it after graduation (57% vs 32%), suggesting that clear permission, rather than prohibition, is what helps students develop a healthier relationship with the technology.
But none of this accounts for whether they actually want to use it in the first place.
They really do hate it
The clearest sign that something has shifted comes from Gallup’s annual survey for the Walton Family Foundation and GSV Ventures, which has tracked Gen Z’s feelings about AI year on year. The generation once described as AI’s natural early adopters has soured on it noticeably; excitement and hope are both falling, anger is climbing, and the emotions now sitting at the top of the list – much like the Australian research suggests – are curiosity and anxiety rather than enthusiasm. Gallup’s researchers tie the rising anger to AI dimming the prospects of entry-level workers, noting that the oldest members of the cohort, the ones closest to the job market, are the angriest of all. When asked whether they would rather a human or an AI handle something that matters to them, like tutoring or advice, the overwhelming majority still choose the human.
The commencement booing is the audible version of that resentment. Students are jeering at a technology they use daily and have come to associate with a thinner education, a narrower set of career prospects, and a creeping doubt about their own competence. The mood is not confined to campuses, either. A widely reported survey from Writer and Workplace Intelligence found that a striking share of Gen Z workers admitted to quietly sabotaging their own employers’ AI rollouts, by refusing mandated tools, feeding junk into them, or letting them fail on purpose, often for the simple reason that they do not want the technology to take their jobs. Through the spring of 2026 the business and tech press, from Fortune to Futurism to Fast Company, settled on a single framing for all of this: that the more Gen Z uses AI, the more they seem to hate it.
Tech critic, author, and inventor of the term “enshittification” Cory Doctorow put a useful frame around this in a piece published the same week as the Arizona graduations. The early web, he argues, was a technology that workers smuggled into their workplaces against the wishes of their IT departments, and that young people pushed their employers to accommodate, because it plainly helped them do their jobs; AI is the mirror image, a technology that employers now have to force on reluctant workers and that the young are actively resisting. He grounds the inversion in the economics underneath it, observing that the web grew more useful and more profitable with every new person who joined, so its adoption could be voluntary and bottom-up, while AI loses money with every new user, so its adoption has to be coercive and top-down. Universities are in the middle of this, and schools only a little way upstream.
What does this mean for schools?
If you are a school leader investing in Copilot licences, rolling out Gemini for Education, or drafting a new AI policy, the evidence above should give you pause, not because bringing GenAI into education is necessarily wrong, but because the ground on which you are building is far less stable than the sales pitch suggests.
Students are not enthusiastic early adopters waiting for institutional permission. They are reluctant, anxious users who already have access to better tools than most schools are offering and who associate AI with labour market threat, academic dishonesty, and cognitive decline.
Framing the discussion as “AI literacy” risks provoking the backlash rather than preventing it. When students hear “AI literacy,” many will hear “learn the technology that is going to replace you.” The commencement booing happened precisely at the moment when authority figures stood up and told young people to “embrace” a technology that threatens their futures. Classroom teachers attempting the same message, however well-intentioned, may face the same reaction. The problem is not just communication but relevance: students do not engage with guidance that does not address their actual concerns.
Assessment redesign, not AI policy, is the real lever, and conflating the two is a mistake. Princeton and Stanford have both reverted to proctored, handwritten exams, and many Australian universities are heading back to supervised, in-person assessment. In higher education, this is a candid admission that existing assessment structures cannot survive contact with GenAI. Schools, though, have advantages that universities lack, since we see our students more often and in smaller cohorts, and K-12 is staffed overwhelmingly by qualified teaching professionals rather than by researchers and subject experts who may never have trained as teachers. The productive question for schools is not “how do we stop students using AI?” but “what forms of assessment remain meaningful when AI is freely available?” That question is uncomfortable because it calls for redesigning tasks rather than purchasing software, and it demands pedagogical expertise rather than technological investment.
The 2026 US graduation season has made visible something the research has been showing for over a year: students are caught in a structural bind where AI use feels simultaneously compulsory and corrosive. They use it because the assessment system incentivises it, because the labour market seems to demand it, and because everyone around them is using it. They resent it because it threatens their careers, undermines their confidence in their own thinking, and turns their education into something that feels increasingly hollow.
Schools responding to this moment with AI policies, tech licences, and AI literacy programs need to understand that they are not introducing a neutral technology into a receptive population. They are intervening in an emotionally charged, structurally complex situation where students’ dominant feelings are anxiety, guilt, and anger, not excitement. Students are telling us – loudly, sometimes aggressively – that their relationship with AI is more complex than “use it or don’t”. We need to listen to them.
Want to learn more about GenAI professional development and advisory services, or just have questions or comments? Get in touch:
Leave a Reply