By Pankaj Doshi

OpenAI recently announced that an internal AI system had produced a proof addressing the Navier-Stokes existence and smoothness problem, one of Mathematics’ seven Millennium Prize Problems. The claim will, as it should, be examined carefully by mathematicians. But whether every detail survives scrutiny is not the most important question for universities today. The more important fact is that AI systems are now operating at a level where they can participate in research that, until very recently, represented the outer edge of human intellectual effort. This should be a call to action for higher education. What should a university teach when increasingly capable machines can execute a growing share of the intellectual work that universities have traditionally trained people to perform? The appropriate response is to raise the level of human inquiry. When calculators became universal, we did not stop teaching mathematics; we stopped treating arithmetic as the summit of mathematical ability. AI demands a similar, but much larger, shift.

Advertisement

Technical depth will remain essential, but technique alone will no longer be enough. A PhD student must learn to identify consequential questions, frame them rigorously, combine knowledge across disciplines, use AI systems intelligently, challenge their outputs, and connect computational answers to physical reality. Doctoral training should ai therefore become more ambitious. Undergraduate education needs an equally serious redesign. Much of science and engineering teaching still revolves around lectures followed by problem sets in which students reproduce standard methods under examination conditions. AI can already solve many such problems convincingly.

Classrooms must move more quickly from “solve this equation” to “decide what should be modelled, what assumptions are defensible, what data are missing, how the answer could fail, and how you would test it in the real world”. This shift also has implications for faculty. The availability of extraordinary intellectual tools should not lead only to faster production of conventional papers. It should enlarge the scale of problems we are willing to attack. Faculty members at India’’s leading institutions — the IITs, IISc, IISERs and major universities — should ask whether research agendas designed for a pre-AI era are ambitious enough for the capabilities now becoming available. Code can be written and experiments planned far faster than before. None of this eliminates the need for scientists and engineers. It changes where their scarce human judgement should be spent: On choosing the right problem, setting constraints, recognising nonsense, designing decisive experiments and understanding consequences.

For institutions, the policy response should be deliberate. AI literacy should become part of the core training of every science and engineering student. If intellectual execution becomes dramatically cheaper and faster, then ambition, problem selection and the ability to translate ideas into reality become even more important sources of advantage. The Navier-Stokes announcement should therefore be read less as a story about a machine competing with mathematicians and more as a warning to universities that the intellectual baseline is moving. The correct response is neither panic nor resistance. It is to demand more from students, faculty and institutions.

Advertisement

The writer is visiting faculty, Department of Chemical Engineering, IIT-Bombay. Views expressed are personal