(Oct 23): Recently, I spoke with the parent of a four-year-old child who made a striking prediction: “By the time my child is ready for university, universities might no longer exist.”

As a lecturer, I found the remark unsettling, but not unreasonable. Parents deserve to ask whether the time, fees and sacrifices involved in higher education will still be worthwhile. Artificial intelligence (AI) can already explain theories, generate essays, analyse financial data and provide personalised tutoring within seconds. If students can access knowledge so easily, why should they spend three or four years pursuing a degree?

I believe universities will still matter in 2040. But institutions that continue to reward memorisation and routine assignments may struggle to justify their relevance. Their future depends on a fundamental shift: from delivering information to developing independent judgment.

A 2025 German experiment by Eichinger and colleagues involving 68 university students compared conventional lectures with ChatGPT-supported self-study. Students using AI completed learning activities faster, with no statistically significant difference in performance on short tests. Yet speed is not the same as lasting understanding.

Another 2025 study by André Barcaui involving 120 undergraduates found that, on a test administered 45 days later, students who had used ChatGPT as a study aid scored 57.5%, compared with 68.5% among students using traditional methods. These studies have different designs and neither settles the debate, but together they raise an important question: does easier access to answers always produce better learning?

Imagine visiting a gym and letting a machine lift every weight. The exercise gets done, but your muscles do not grow by watching. Similarly, students who outsource every argument or calculation to AI may finish assignments without developing the reasoning those tasks were meant to teach.

That does not mean AI should be banned. A 2026 study involving 100 undergraduates at Universiti Malaysia Sarawak reported a positive association between AI reliance and self-reported higher-order thinking engagement. The issue is not simply whether students use AI, but how thoughtfully they use it.

Consider an accounting student in 2040. AI may prepare financial statements, assess sustainability disclosures and draft audit reports in minutes. But can the graduate identify manipulated figures, question misleading environmental claims or exercise professional scepticism when an AI-generated conclusion looks convincing yet is wrong?

An engineering graduate may need to challenge the safety assumptions behind an AI-designed bridge. A medical graduate must weigh an AI-assisted diagnosis against a patient's circumstances. In each case, technology can recommend a course of action; a qualified professional must still evaluate the evidence, defend a decision and accept responsibility for its consequences.

That is what a degree should ultimately certify: not merely technical proficiency, but the ability to make sound decisions under uncertainty. Employers need graduates who can explain their reasoning, recognise their own limitations and uphold professional standards when a convenient answer is not a responsible one.

University assessments should reflect this reality. Instead of asking students only to produce reports that AI can readily generate, lecturers could present an AI-generated financial analysis containing flawed assumptions and ethical dilemmas. Students would identify weaknesses, propose alternatives and defend their conclusions before a panel.

Structured debates, real-world cases and collaborative projects could complement such assessments. AI literacy should also be taught alongside disciplinary knowledge, helping students recognise bias, verify sources and understand the limits of automated systems. The aim is to teach students to think with AI without losing the ability to think independently.

Universities also offer something that cannot be measured by the speed of an explanation. A 2024 survey by Chan and Tsi of 399 students and 184 teachers across eight Hong Kong institutions highlighted the continuing importance participants placed on human educators, particularly for emotional understanding and personal development.

A lecturer can encourage a student facing failure, challenge assumptions and guide difficult ethical conversations. On campus, students learn to disagree respectfully and collaborate with people whose experiences differ from their own. These relationships remain valuable even as AI becomes a more capable learning companion.

Research is equally important. Universities do not merely transmit existing knowledge; they test hypotheses, conduct experiments and investigate unfamiliar problems. AI can accelerate discovery through modelling, data analysis and new hypotheses, but findings still require rigorous validation, interpretation and consideration of social consequences. Researchers must also decide which questions are worth asking, whose needs are overlooked and how discoveries should be used.

Nor should technological progress widen educational divides. AI tutoring may improve access, yet students without reliable internet, strong foundations or digital skills could fall further behind. Universities must continue providing inclusive mentorship, research facilities and opportunities that technology alone cannot guarantee.

Malaysia's Higher Education Blueprint 2026–2035 emphasises human-centred education, adaptability, ethical values and AI integration. These ambitions are welcome, but their success will be determined in classrooms, not policy documents. 

Investing in AI platforms alone is not educational reform.

Are universities redesigning examinations to reward reasoning rather than reproduction? Are lecturers receiving the training and freedom to adopt new teaching methods? Are graduates evaluated on their ability to solve ambiguous problems and act responsibly, rather than grades alone?

By 2040, universities may offer more flexible degrees, shorter professional credentials and lifelong learning, while AI handles routine tutoring and administration. Lecturers could devote more attention to mentorship, discussion and research.

When that four-year-old child reaches university age, I hope the institution they enter does more than teach them to obtain the right answer. It should teach them which answers to question, which decisions to defend and which values must not be compromised.

Perhaps the greatest threat to universities is not that AI will replace them, but that they may fail to reinvent themselves before students no longer see the value of attending.

Dr Najihah Abd Razak is a lecturer in accounting and finance at the School of Accounting and Finance, Asia Pacific University.