Artificial intelligence is forcing universities to reconsider what a degree should prove. The immediate dispute concerns whether students may use systems such as ChatGPT in assessed work. The larger question is how higher education can prepare graduates for jobs in which AI may handle routine drafting, analysis and coding, while human judgment remains essential.

A recent briefing from the UK Parliament’s House of Lords Library said a 2026 student survey found that 95% of respondents had used AI in their studies. Ninety-four percent reported using it for assessed work, while 12% said they had used AI-generated text directly. The figures indicate that blanket bans may be difficult to enforce, even as universities remain responsible for ensuring that submitted work represents genuine learning. The Lords Library briefing also describes a shift towards assessments that either explicitly permit AI or use formats less vulnerable to undisclosed assistance.

From policing tools to testing judgment

Some institutions are responding by changing the assessment itself. The University of Surrey says that, from 2026, it will embed discipline-specific AI capability across every degree, with students expected to understand the technology’s limits, risks and professional consequences. Its planned assessment changes emphasise process, decision-making and applied problem-solving rather than relying mainly on a final written product. The university’s announcement presents the approach as an effort to make graduates’ abilities more demonstrable and trusted.

Other proposals include oral examinations, reflective accounts explaining how AI was used, collaborative projects and real-world tasks. A University of Manchester study, reported by Phys.org, argues that these methods can test understanding, reasoning and judgment more directly than conventional essays.

That does not settle the fairness question. Oral exams and supervised tasks can make authorship clearer, but they may require more staff time and could disadvantage students with anxiety, disabilities or limited access to quiet study spaces. AI-permitted assignments may reflect workplace practice more closely, yet they can make it harder to distinguish a student’s own reasoning from the quality of the tool they used.

The first job is changing too

The pressure on universities is amplified by a weaker graduate labour market. A 2026 analysis from Prospects, based on more than 180,000 UK-domiciled first-degree graduates who completed university in 2023-24, says outcomes have weakened since the post-pandemic hiring boom. It reports pressure in some occupations, including software development, management consultancy and graphic design, but says there is little evidence of wholesale replacement across the graduate labour market. The analysis also notes growth in accountancy, financial analysis and legal professions—fields where AI is expected to change work rather than eliminate the need for graduates.

That distinction matters for students deciding whether university remains worthwhile. A degree may no longer guarantee a smooth transition into professional employment, but employers may increasingly value graduates who can use AI without accepting its outputs uncritically. The Lords Library cited evidence that entry-level jobs exposed to AI are beginning to demand capabilities once associated with more senior roles, including strategic thinking and leadership.

There are signs that the transition is uneven. The Institute of Student Employers says the UK economy remains weak and graduate recruitment is likely to reflect that environment in 2026, with higher early unemployment and more graduates entering postgraduate study. In Canada, a report discussed by The Hub found that 30% of unemployed university graduates aged 20 to 24 had been searching for work for at least six months in early 2026. The employers’ analysis and the Canadian report point to a broader problem: the qualification may remain valuable, while the first rung of the career ladder becomes harder to reach.

A test of what universities promise

The emerging consensus is not that students should avoid AI, nor that universities can simply permit it everywhere. Instead, institutions are being pushed to show that graduates can define a problem, use tools responsibly, verify evidence, explain decisions and accept accountability for the result.

For students, that could mean more transparent rules and assessments that resemble professional work. It could also mean greater pressure to develop skills beyond producing polished text. For universities, the challenge is to provide those opportunities equitably while preserving the credibility of degrees.

AI may reduce the value of some routine tasks, but it has not removed the need for educated workers. The institutions best prepared for the future of work will be those that assess not merely what students submit, but how they reason, what they can defend and whether they understand the consequences of their choices.

Sources