A student in 1995 needed a library to access a journal article. A student in 2010 needed a laptop. A student in 2026 needs a prompt. The information that universities once monopolised is now available to anyone with a smartphone and an internet connection, delivered instantly, in any language, at any level of complexity.

That shift has forced a question that higher education has avoided for years: if information is free, what exactly are students paying for?

The Old Bargain Has Collapsed

For most of the twentieth century, the university's value proposition was straightforward. It was a repository of knowledge. It held the books, the journals, the laboratories, the experts. To learn, you had to go where the knowledge was.

That monopoly is gone. AI can summarise a textbook in seconds. It can explain quantum mechanics to a twelve-year-old or a graduate student. It can generate essays, solve equations, write code, and translate between languages. Whatever can be reduced to information can now be obtained without a lecture hall, a semester, or a fee.

If universities continue to position themselves primarily as information providers, they will lose. They cannot compete on speed, cost, or convenience. They will be undercut by tools that never sleep and never charge tuition.

The question is not whether universities survive. It is whether they can articulate what they still do that AI cannot.

What AI Cannot Do

There is a list, and it is longer than the panic suggests.

AI cannot teach a student to doubt it. It cannot cultivate the instinct to ask whether a confident answer is actually a correct one. It cannot model the slow, uncomfortable process of changing your mind when the evidence demands it. It cannot sit with a student through the frustration of a problem that refuses to yield, and teach them that the frustration itself is part of learning.

These are not soft skills. They are the core competencies of an educated person. And they are precisely what a chatbot, by its nature, cannot supply — because a chatbot is designed to produce answers, not to interrogate them.

A university that teaches students to use AI critically — to check its outputs, challenge its assumptions, and recognise its limits — is performing a function that no AI can replicate. It is training the scepticism that keeps the tool honest.

The Mentorship Gap

There is also something that happens between a student and a teacher that cannot be digitised.

A lecturer who has spent thirty years in a field knows which questions matter and which are distractions. They know where the literature is weak, where the consensus is fragile, and where the next breakthrough is likely to come from. They can look at a student's work and see not just errors, but habits of mind that need correcting.

AI can assess grammar. It can flag factual inconsistencies. It cannot tell a student that they are asking the wrong question. It cannot recognise the moment when a struggling undergraduate is on the verge of a genuine insight and needs only a nudge, not an answer.

That relationship — the apprenticeship of thought — is the university's oldest and most durable function. It is also the hardest to scale, which is why it has been neglected in favour of mass lectures and online modules. The AI era makes it urgent again.

The Credential Question

Employers do not hire graduates because they possess information. They hire them because a degree signals something: that the holder can complete a long project, meet deadlines, work with others, and absorb complex material. The degree is a proxy for reliability and capability.

AI threatens that proxy. If a student can generate a passable essay with a prompt, the essay no longer demonstrates anything. If an examination can be answered by a machine, the examination is worthless. The entire assessment architecture of higher education assumes that the work submitted reflects the abilities of the person submitting it. That assumption is no longer safe.

Universities that fail to adapt their assessments will produce graduates whose credentials mean less every year. Employers will notice. They will start hiring on the basis of portfolios, tests, and interviews instead — and the degree will lose its currency.

The institutions that survive will be the ones that redesign assessment around what cannot be outsourced: oral defence, live problem-solving, collaborative projects, and supervised research.

Malaysia's Specific Problem

Malaysia produces roughly 232,000 graduates a year into a job market with about 127,000 positions that match their qualifications. Nearly two million Malaysians — about 35 per cent of employed degree and diploma holders — work in jobs that do not match their qualifications. Over 65 percent of fresh graduates start their careers earning less than RM3,000 a month.

Those figures existed before AI. But AI accelerates the problem. Since 2020, artificial intelligence has been linked to job losses for close to 300,000 workers in Malaysia. Entry-level roles — the first rung of the career ladder for graduates — are precisely the roles most vulnerable to automation.

Malaysian universities cannot afford to respond by doing what they have always done. The pipeline that produced hundreds of thousands of graduates for jobs that no longer exist is not a pipeline that can be fixed with cosmetic reform.

What Universities Must Do

The path forward is not mysterious, but it is demanding.

Teach thinking, not content. Content is free. Judgment is not. A curriculum that tests recall is a curriculum that AI has already made obsolete. A curriculum that tests reasoning, synthesis, and the ability to identify weak arguments is one that AI cannot replace.

Integrate AI rather than ban it. Prohibiting AI tools is not a strategy. Students will use them regardless. The task is to teach students to use them well — to prompt carefully, verify ruthlessly, and recognise when the tool is wrong. A graduate who cannot use AI critically is as disadvantaged as a graduate who cannot use a spreadsheet.

Restore the tutorial. The mass lecture was always a compromise. AI makes it untenable. Small-group teaching, where students argue, defend, and revise their thinking under expert supervision, is the format that delivers what students are actually paying for.

Rebuild assessment. If the credential is to mean anything, it must be earned under conditions that demonstrate genuine capability. That means more oral examinations, more live work, more supervision, and fewer take-home essays that no one can verify.

The Reckoning Is Not a Threat

The AI era is not the first time universities have been forced to justify their existence. They survived the printing press, the public library, the television, and the internet. Each time, the doomsayers predicted collapse, and each time, the institutions adapted.

But adaptation requires honesty. It requires universities to stop defending the lecture hall as though it were sacred and start asking what students actually need. It requires them to stop treating assessment as an administrative burden and start treating it as the foundation of their credibility. It requires them to recognise that their monopoly on information is gone, and that their future lies in something harder to replicate.

The information is free. The judgment is not. Universities that understand the difference will thrive. Those that do not will find themselves competing with a chatbot — and losing.