If AI becomes the primary tutor, what is the academic actually for

The assumption that no longer holds

Higher education has long operated on the premise that learning happens primarily through expert-led instruction. That premise is now being tested by systems that can do much of the same work, often more responsively.

Knowledge delivery was never the whole job, but it dominated the timetable

Consider the typical structure of a taught master's programme. A significant proportion of contact time is spent on lectures, seminars, and guided reading, activities centred on explaining concepts, answering questions, and walking learners through material. Assessment and deeper intellectual engagement often occupy a much smaller share of an academic's week than preparation and delivery.

AI tutoring systems are now performing many of these functions competently. They explain concepts at the learner's pace. They offer instant formative feedback. They adapt to individual gaps. They are available at any hour, in any time zone. They do not tire, and they do not have 200 other students competing for their attention.

This is not a speculative scenario. Institutions are already deploying AI-driven tutoring at scale. The London School of Innovation, for example, has built its entire learning model around a proprietary virtual tutor that delivers personalised guidance, formative assessment, and adaptive learning paths, with human academics deliberately repositioned into different functions.

What changes when explanation becomes automated

If AI handles the bulk of knowledge transmission, a set of consequences follows. The academic's scarcity value no longer lies in what they know, because that knowledge is increasingly encoded in systems that can deliver it more efficiently. Nor does it lie in availability, because AI does not have office hours. The question then becomes: what can an academic do that an AI system, however sophisticated, cannot?

Intellectual arbitration as a reframed purpose

One compelling answer is that academics become judges of reasoning quality rather than providers of instruction. This is not a lesser role. It may be a harder one.

From instructor to arbiter

There is a meaningful difference between explaining a concept and evaluating whether someone has genuinely understood it. AI can assess whether an answer matches a known pattern. It is far less reliable at judging the quality of reasoning in ambiguous, contested, or novel domains. Consider a postgraduate student analysing the ethical implications of deploying facial recognition in public healthcare. An AI tutor might scaffold the analysis and provide relevant frameworks. But determining whether the student's reasoning reflects genuine intellectual maturity, whether they have weighed competing considerations with appropriate nuance, whether their conclusions would hold under cross-examination, these are judgements that depend on epistemic standards, disciplinary expertise, and contextual sensitivity that current AI systems do not possess.

This reframing positions the academic as an intellectual arbiter: someone who validates not just correctness, but the quality of thinking. It is closer to the role of a doctoral supervisor or a professional examiner than a classroom instructor.

Capability validation, not content coverage

If this model gains traction, the academic's contribution shifts decisively towards what might be called capability validation. The focus moves from whether content has been covered to whether a learner can reason, judge, and perform in conditions of complexity. This has significant implications for how academic workload is structured, how performance is evaluated, and what hiring criteria look like. An institution that embraces this model needs academics who are excellent at assessing reasoning under uncertainty, not simply experts in delivering polished lectures.

Science of Adult Education

Traditional educational paradigms often struggle to adapt to rapidly advancing technologies. At the intersection of learning science and AI, this module stands as a beacon for educators seeking to navigate and flourish in this evolving environment....

Learn more

Two possible futures, shaped by decisions made now

There is no single trajectory here. How institutions respond to AI tutoring will likely produce divergent models of the academic role, each with different implications for quality, cost, and institutional identity.

Possibility one: the academic as quality controller

In one scenario, AI handles the majority of instruction and formative feedback, while academics concentrate on summative assessment, research supervision, and the interpretation of complex problems. Their role becomes analogous to an auditor or a judge: intervening at critical decision points rather than managing the day-to-day flow of learning. This model could dramatically improve the ratio of meaningful intellectual engagement to routine delivery. It could also make the academic role more satisfying, particularly for those who entered the profession to advance knowledge rather than repeat introductory material year after year.

Possibility two: the academic as curriculum architect

In another scenario, the primary value of the academic shifts upstream, towards designing the learning experiences, simulations, and assessment frameworks that AI systems then deliver and support. Here, the academic is less a judge and more an engineer of intellectual challenge. They design the scenarios that test real-world decision-making. They determine what counts as mastery. They set the epistemic standards that AI systems enforce but cannot originate.

Both futures are plausible. Both require institutions to make deliberate choices about academic identity, workload models, and governance. Neither emerges automatically from purchasing an AI platform.

The risk of inaction

A third possibility, and the least productive one, is that institutions simply layer AI tools on top of existing structures without rethinking the academic role at all. In this scenario, AI tutors coexist awkwardly with traditional lectures, creating redundancy rather than redesign. Students receive conflicting signals about where authority lies. Academics feel threatened rather than liberated. This is not a technology problem. It is a leadership and governance problem.

What empirical clarity the sector still needs

Much of the current debate about AI in education is driven by intuition rather than evidence. A few carefully scoped studies could move the sector towards significantly greater clarity.

Measuring reasoning quality across delivery models

One area ripe for rigorous investigation is comparative assessment of reasoning quality. If two cohorts study the same material, one primarily through AI-driven tutoring and the other through traditional academic instruction, and both are then assessed by independent human examiners on the quality of their reasoning in complex, ambiguous scenarios, what differences emerge? This is not a study of content recall or exam scores. It is a study of whether delivery mode affects the depth and robustness of intellectual capability. The methodological challenges are real, but a well-designed multi-institutional study, ideally across different disciplines and cultural contexts, could provide evidence that currently does not exist. It would help answer whether AI tutoring produces learners who know things, or learners who can think through things, and whether there is a measurable difference between those outcomes.

Such a study would be valuable precisely because it resists easy answers. It might reveal that AI-tutored learners outperform in some dimensions and underperform in others. It might show that the combination of AI instruction with human intellectual arbitration produces the strongest outcomes. Or it might challenge assumptions on all sides. Whatever the findings, the sector would benefit from evidence rather than ideology.

A decision test for the decade ahead

The question facing institutional leadership is not whether AI will change the academic role, but whether the redesign will be deliberate or accidental.

Redesign as opportunity

There is a genuine opportunity here to elevate the academic profession. If routine instruction migrates to AI, academics could spend more of their time on the work that drew them to scholarship in the first place: pushing the boundaries of knowledge, mentoring advanced thinking, and holding learners to the highest standards of intellectual rigour. This is not a diminished role. It is a more focused one.

The institutions that thrive will be those that articulate clearly what their academics are for in an AI-augmented environment, and restructure accordingly. That means revisiting promotion criteria, workload models, hiring profiles, and the implicit contract between institution and faculty. It means treating AI not as a threat to be managed but as a catalyst for institutional clarity.

The decision test is straightforward. If an AI system can already perform 60% of what a given academic role currently involves, is the institution investing in making the remaining 40% extraordinary? Or is it defending the full 100% on the grounds that it has always been done that way?

And the uncomfortable question that follows: if the most valuable thing an academic can do is judge the quality of someone else's thinking, how confident is each institution that its current academics were hired, developed, and rewarded for precisely that capability?