Beyond Sanyal's Sensationism: Why India Must Reform, Reinvent, and Expand the University in the AI Era

Part I: The University Is Being Put on Trial

There is something peculiar about the current debate on Indian higher education. Almost everyone agrees that India's universities need reform, and that proposition is hardly controversial anymore: curricula in many institutions lag behind technological and economic change, teaching quality varies enormously, industry linkages remain weak, research output is uneven, many graduates require substantial workplace training, and the adoption of artificial intelligence across teaching, research and administration remains far below what the technology now makes possible. These are legitimate concerns.

But a more radical proposition has begun to emerge from some quarters: that the university itself is becoming obsolete. Artificial intelligence can explain concepts, generate essays, write code, analyse information, and tutor students. Online platforms can provide lectures from leading academicians anywhere in the world. Employers increasingly complain that conventional degrees do not guarantee workplace capability. Startups can build highly skilled teams without reproducing the structures of traditional universities. And some academic disciplines appear increasingly disconnected from the economic and technological realities around them. 

From these observations, a seductive conclusion follows: why maintain expensive physical universities at all, rather than shifting education online, using AI for personalised instruction, and reserving physical institutions for practical training, research and perhaps entrepreneurship?

Sanjeev Sanyal, a member of the Economic Advisory Council to the Prime Minister, has articulated this argument repeatedly. In his latest podcast-interview to Smita Prakash, on 19 August, he questioned the continuing relevance of conventional university education, highlighted the ability of AI to provide instruction and assessment, questioned the occupational value of some humanities degrees, and pointed to IIT Madras as an example of what universities should increasingly become — intellectually active ecosystems combining research, practical work, entrepreneurship, and industry.

There is much in this diagnosis that deserves serious consideration, and much that does not. The problem is not that Sanyal is asking uncomfortable questions — India needs people willing to ask uncomfortable questions about its universities. The problem begins where legitimate criticism of an institutional model gets converted into sweeping conclusions about the institution itself. That is not radical reform; it is intellectual overreach.

1.1 Naming what is actually happening

What is underway in this discourse is not really a debate about curricula or pedagogy. It is an erosion of legitimacy — the cumulative effect, across podcasts, columns, and advisory-adjacent commentary, of treating the university as an institution whose burden of proof has quietly shifted from "how should it be reformed" to "why should it continue to exist at all." That shift matters, because once legitimacy itself becomes the question, the answer tends to arrive as a single metric, a single failure, or a single successful exception mistaken for a rule. The university is a much broader and deeper institution than any legitimacy arc built on employability statistics, lecture formats, or one exceptional campus can capture. Judging it on one axis and pronouncing on the whole is not scrutiny; it is a category error dressed up as candour.

It is also worth noticing who is not participating in this legitimacy arc. In the same week the Prakash/Sanyal podcast gained popularity, the Prime Minister, on 20 August, chaired the third in a series of high-level meetings with GoI Secretaries — this one on Infrastructure and Connectivity, Security and External Affairs, and Governance. Among his stated priorities was deeper engagement between government departments and universities and young people, specifically to bring fresh, unconventional thinking into policymaking. This was not a speech pitched at public opinion; it was an internal instruction to the government's own bureaucracy, delivered to Secretaries rather than to Ministers, about where fresh thinking should come from. That is a difficult fact for the obsolescence narrative to absorb. The apex of India's executive is, in the same week this argument is advancing, leaning on the university more, not less. An institution as spent as the legitimacy arc suggests - would be a strange one for the state itself to be depending on.

1.2 The old university does need to change

Let's start by granting the strongest version of the argument. The traditional university emerged in conditions of information scarcity — knowledge concentrated in books, libraries, professors, and physical institutions, and a student travelled to a university because that was where the knowledge was. That world has changed. A student with a smartphone can access lectures from outstanding universities anywhere; digital libraries have dramatically expanded access to scholarly material; online courses can teach programming, mathematics, finance, languages and dozens of professional subjects at low cost; and generative AI can explain on demand, generate examples at varying difficulty, simulate conversations, critique drafts and walk students through problems.

The lecture can no longer be treated as the unquestioned centre of university education, and neither can the conventional take-home assignment: if an AI system can produce a competent essay, write a substantial piece of software, or solve a standard problem set, the educational value of simply asking students to submit such outputs has to be reconsidered. Universities will have to rethink assessment toward demonstrations, oral examinations, supervised work, practical projects, research experience, and collaborative exercises — methods that establish what a student actually understands and can do — and they will have to teach students to work with AI rather than pretend it does not exist. On all these, the critics are right.

But there is an enormous difference between saying the lecture is becoming less important and saying the university is becoming less important. The first is an educational observation. The second is an institutional proposition that requires considerably more evidence — and it is exactly that evidentiary gap the argument keeps skipping past.

1.3 The category errors begin when everything becomes "skills"

One of the most persistent problems in this debate is the tendency to collapse several distinct things into a single category called "skills". A degree is not a skill. A discipline is not a job description. A curriculum is not an occupation. A university is not a corporate training department. And employability is not synonymous with education.

A mechanical-engineering graduate may need additional training before becoming a productive industrial engineer — that does not mean the degree was pointless. A biotechnology graduate may need to learn the procedures of a specific regulated manufacturing facility — that does not mean molecular biology should have been replaced by a company-specific training module. The same logic runs through computer science, business, economics, law, sociology, etc: a university cannot reproduce every future workplace, because the workplace itself keeps changing. The more useful question is not whether a degree makes a graduate immediately employable, but whether the education makes the graduate capable of becoming productive in a changing professional environment — a distinction that will matter more, not less, as AI reshapes the composition of work.

1.4 The university is not supposed to manufacture finished employees

There is an understandable desire to make universities more responsive to employers, but responsiveness can easily tip into subordination. If every curriculum is redesigned around today's precise employer requirements, universities will spend their existence chasing yesterday's economy — and a four-year degree-program cannot sensibly be rewritten every six months because an occupation has changed. The proper function of higher education is to provide foundations deep enough that graduates can keep learning as the environment changes: disciplinary knowledge, analytical ability, communication, digital capability, curiosity, problem-solving, and the capacity to acquire further knowledge on their own. That makes a graduate trainable. The workplace then supplies the specific operational experience a profession requires.

This distinction matters because the first job has historically performed a function that discussions of employability tend to ignore: it has been a training institution in its own right. A young engineer, analyst, programmer or scientist enters an organisation with limited practical experience, does supervised work, makes mistakes, runs into real constraints, and gradually develops professional judgement. The first job is not merely a destination after education — it is part of the education-to-experience system. Artificial intelligence may well disrupt that mechanism, and that is a serious problem, but the answer is not to force universities to impersonate workplaces. It is to build a better bridge between the two.

1.5 AI does not eliminate the institutional value of universities

A recent report from MIT on this point is instructive. Its response to AI's effect on teaching, assessment, and research training has not been to declare the university obsolete but to rethink what happens inside it. If AI can produce a first draft, students may need to defend and critique their own work. If AI can write code, they may need to demonstrate they understand the underlying architecture and can evaluate what the AI generated. If AI can tutor, professors can spend more time on discussion, experimentation, and research. The implication is not that human education disappears — it is that the scarce educational resource changes. When information becomes abundant, judgement becomes more valuable. When routine cognitive production gets cheap, the ability to ask good questions becomes more valuable. When AI can generate plausible answers, recognising an implausible one becomes more valuable. When machines can produce increasingly sophisticated outputs, the capacity to take responsibility for consequential decisions becomes more valuable — and these are precisely the capabilities universities can cultivate through sustained intellectual and practical interaction.

This is why the physical university should not be reduced to a building full of lecture rooms. Its value lies in the ecosystem: students encountering professors, researchers encountering students, engineers encountering scientists, entrepreneurs encountering researchers, industry and government problems entering academic environments, experiments happening in laboratories, teams forming, arguments happening around tables and boards, ideas circulating formally and informally. AI can augment this ecosystem. It cannot automatically reproduce it.

1.6 IIT Madras is an example — not a universal template

The IIT Madras ecosystem — its combination of advanced research, startups, laboratories, engineering projects, and industry engagement — is genuinely impressive, and represents an important model for the future. But an example of a successful configuration should not automatically become a prescription for the entire system. 

India has thousands of higher-education institutions operating in radically different contexts: some concentrated in industrial clusters focusing advanced manufacturing, materials or process engineering; some connected deeply to agriculture and allied sectors; some specialised in medical sciences and services; some serving as a national research centre in a strategic technology; some serving Himalayan regions through ecology, hydrology, and disaster management research and education; others organised as federations of public colleges providing multidisciplinary education across a district or division; etc. There is no reason for all of them to become startup incubators — trying to remake every university as an IIT Madras-style startup ecosystem would itself be a kind of institutional monoculture. India does not need every university doing the same thing. It needs the system as a whole doing many more things.

It is worth being explicit about what this means for the institution Sanyal singled out for praise. IIT Madras is not a counter-example to the university as an institution — it is the university, functioning precisely as it should when its incentives, culture, and connections are right. Holding it up as proof that universities in general should shrink -  comes close to self-refutation: the exhibit for the prosecution turns out to be a witness for the defence.

1.7 The missing distinction between a discipline and an ideology

The same problem becomes sharper when the discussion turns to the humanities and social sciences. There are legitimate criticisms here too. Swathes of these disciplines, across secular democracies, have drifted into leftist/Hinduphobic activism, intellectual homogeneity, and weak engagement with empirical and contemporary problems; too many practitioners have stopped testing inherited assumptions against reality, and ideological conformity has substituted for intellectual competition. This deserves criticism.

But attacking a discipline because some of its practitioners have adopted an ideology is a category error. Political ideology is not an academic discipline. Sociology is not socialism. Economics is not capitalism. History is not Hinduphobia. English studies is not one political worldview either. A discipline's intellectual health has to be judged by the quality of its questions, evidence, methods, arguments, and discoveries — not by the politics of some of its practitioners. And the standard has to be applied consistently: if some sociologists become ideologically closed, reform sociology; if some historians do, reform history; if some economists do, reform economics. The measure is the same in every case — quality of evidence and argument, not proximity to markets or machines. A discipline captured by ideology needs renewal. A discipline that happens to sit closer to industry is not, for that reason alone, healthier. 

The appropriate response to intellectual stagnation is intellectual renewal, not disciplinary abolition — and AI can actually hand these disciplines real instruments for that renewal: sociologists working with large administrative and behavioural datasets, historians digitising and analysing archives, linguists working with multilingual computational systems, economists running more sophisticated analyses and modelling. The task is to put these disciplines into an upward spiral of research, empirical engagement, technological augmentation, and real-world problem-solving — which is a very different project from preserving every existing academic culture unchanged.

1.8 Reform is necessary. Abandonment is not.

The strongest case against India's existing university system is not that universities have become unnecessary. It is that India has not come close to extracting their full potential. The university remains too often confined to a narrow loop — classroom, examination, degree — when it should operate as a much larger one: education, research, experimentation, external engagement, application, knowledge circulation, teaching, student capability, new research.

The debate needs to move past the binary of university versus online platform. Online education will keep growing. AI is already transforming teaching. Many university courses genuinely need redesigning. None of that is in dispute. The real question is what kind of university India needs when intelligence itself is becoming computationally augmented — and that question cannot be answered by telling students to watch more videos, ask AI more questions, take more corporate courses, or build more startups - any more than it can be answered by defending the university exactly as it exists today. Both positions fall short. 

What India needs is more ambitious than either: a university that is more technologically capable, more intellectually open, more connected to industry, and government, more deeply embedded in its regions, more useful to society, more engaged in research, and more capable of turning knowledge into national capability. The university of the future does not need to shrink because AI has arrived. It needs to grow in purpose. Before deciding India has too many universities, it is worth asking whether India has been asking too little of them — and whether legitimacy was ever the right question to begin with.

Part II: The AI-Era University — From Lecture Hall to Knowledge Ecosystem

It would be a mistake to respond to this technological shift by asking only how universities should teach AI. That question is already too small. The deeper question is how a university should function once artificial intelligence has become part of the infrastructure through which knowledge is produced, analysed, communicated and applied.

For generations, universities operated within a relatively stable knowledge cycle: professors acquired knowledge through research and scholarship, transmitted it to students through lectures and texts, assessed what had been learned, and sent graduates into the wider economy. AI is disturbing every part of that cycle — retrieving and synthesising information, generating explanations, writing and debugging code, analysing data, assisting literature reviews, generating simulations, translating across languages, and increasingly performing parts of the research and administrative work once done by academics and staff. The right response is not to bolt an "AI course" onto the existing university degree-programs. It is to make the university itself AI-augmented.

2.1 From AI as a subject to AI as disciplinary infrastructure

There is a real difference between teaching AI as a specialised subject and integrating it into the practice of every discipline that can meaningfully use it. A university can offer a generic machine-learning course and leave most departments intellectually untouched by AI — that is not transformation. The more consequential model is one that runs through every field, as in: physics plus AI, chemistry plus AI, biology plus AI, agriculture plus AI, civil-engineering plus AI, economics plus AI, business plus AI, sociology plus AI, history plus AI, and so on. The goal is not to turn every student into a programmer, but to ensure students understand how computational intelligence changes the way their own discipline generates, tests, interprets, and applies knowledge.

Thus, eg, in mechanical engineering, this might mean working with sensor data and AI-assisted predictive maintenance; in agricultural science, machine learning for crop and yield forecasting, disease detection and climate modelling; in chemistry, AI-assisted molecular discovery, materials design and simulation; in biology, transformed genomics, protein analysis, and drug discovery; i economics, richer empirical analysis and simulation, alongside sharper thinking about causality, data quality and model assumptions; in sociology, analysis of large-scale textual, demographic, behavioural, and administrative data, alongside harder questions about sampling, bias, privacy, and meaning. In history, digitised archives and computational methods opening new possibilities for textual analysis, manuscript preservation, and historical reconstruction; etc. 

The underlying principle is simple: AI should augment disciplinary depth, not replace it. A student who knows how to ask an AI system a question is not, on that basis alone, educated. A student who understands a discipline deeply enough to know what question to ask, what evidence matters, what an answer actually means, when an AI-generated answer is wrong, and what to do about it — that student possesses something far more valuable. Disciplinary education does not disappear in the AI era. It becomes more important.

2.2 The professor cannot remain outside the transformation

There is an uncomfortable implication here. Curriculum reform alone will not produce this transformation, because courses are downstream of faculty behaviour. A university can mandate "AI integration" across every syllabus, introduce electives, redesign course descriptions and announce digital initiatives — but if faculty keep researching, teaching, assessing, and administering essentially as they did before generative AI arrived, the transformation will stay cosmetic.

The professor becomes the critical point of reform. The academician of the AI era cannot remain merely a transmitter of inherited knowledge — not by abandoning scholarship or becoming a corporate trainer, but by becoming more permeable to the world outside the university — by functioning as an knowledge node of industry, government, research institutions, professional bodies, startups, and philanthropy - wherever relevant to the discipline. An industrial engineer should understand contemporary industry's operational problems. An environmental scientist should encounter actual ecological systems and public programmes. An economist should engage real businesses, financial institutions, government data and changing economic structures. A labour sociologist should not build an intellectual universe entirely out of old theoretical debates while ignoring how automation, platforms, migration and AI are reshaping work. A historian should be able to engage digitisation, archives, museums and heritage institutions. None of this is about commercialisation. It is about intellectual circulation.

2.3 Engagement is not a distraction from scholarship

There is a persistent academic fear that external engagement contaminates scholarship, and it can, if badly designed: a professor dependent on a company for research conclusions can lose intellectual independence; a university that lets political authorities dictate its findings stops performing its public function; consultancy driven purely by revenue can degrade into report production. But this is an argument for institutional safeguards, not for academic isolation. Properly structured, external engagement strengthens scholarship — industry exposes academics to operational realities textbooks rarely capture, government exposes them to problems of scale and implementation, communities expose them to lived realities administrative data can obscure, research organisations expose them to emerging frontiers, philanthropy exposes them to problems in social development and heritage management. The academician then brings those encounters back into the university: a real industrial problem becomes a research question, a government implementation bottleneck becomes a case study, a field observation becomes a dataset, a new technology reshapes a curriculum, a student project generates a hypothesis, and research eventually returns a solution to industry or government. That is a feedback loop. Consultancy, understood this way, is not an activity external to scholarship but an epistemic feedback mechanism — provided intellectual independence and academic standards are protected. The university should not merely export knowledge. It should also import reality.

2.4 The student must remain the ultimate test

There is an important safeguard here, though. External engagement is not valuable simply because a professor has accumulated corporate clients, research projects, or expert committees. The question that matters is what the student has gained. A professor's industry engagement should feed back into the classroom as contemporary case material, real datasets, current technological practice, practical problems, updated reading, applied projects, research opportunities, internships and exposure to professional environments. A government collaboration should similarly open up genuine understanding of how public systems work; a research partnership should give students access to real scientific problems rather than another theoretical exercise. External engagement becomes legitimate, in other words, when it strengthens student capability — external engagement feeding updated teaching, feeding student capability, feeding workplace, or public experience - feeding back into new engagement. Break that loop and the professor becomes a lecturer in morning consultant by afternoon - which is not transformation. The point is to make the university's different activities reinforce each other.

2.5 The university should not fear AI's challenge to faculty

There is a harder question universities should confront directly. If AI can perform parts of teaching, assessment, research assistance, and administration, some existing academic practices will become less valuable — and that is not necessarily a disaster. Universities should not preserve low-value academic labour merely because it has historically existed. 

If AI can automate routine teaching material, professors can spend more time on mentoring, research, discussion and field engagement. If it can provide basic explanations, classroom time can shift toward argument, experimentation and problem-solving. If it can assist literature searches, researchers can spend more time evaluating evidence and designing better questions. If it can automate administrative work, staff can spend more time with students and research. 

The objective should be augmentation, not artificial preservation of obsolete tasks — but augmentation requires capability, and a professor who refuses to understand the tools transforming their own discipline cannot effectively teach students to use them. This is why faculty development has to become central to India's university reform.

2.6 A national faculty-AI capability programme

India should consider structured in-service courses/programs through which (willing) faculty can become AI-augmented academicians - built in at least two layers. 

A generic layer would cover AI-assisted teaching, research assistance, assessment design, academic administration, literature discovery, data analysis, academic integrity, and AI governance and responsible use. 

A discipline-specific layer would recognise that, eg, a chemistry professor needs different AI capabilities from a history professor, a civil engineering professor needs different tools from an economis professor, an agricultural science professor needs to analyse different datasets from a medical science professor, etc.

These courses should be coordinated and developed ideally by representative associations and disciplinary societies. The Association of Indian Universities could potentially coordinate this, working with universities, disciplinary societies, technology organisations and relevant public agencies. The point is not to certify faculty for the sake of another certificate. It is to change academic practice, and the relevant question after any such programme should simply be: what can this professor now do that they could not do before?

2.7 Incentives must eventually follow

Faculty capability cannot depend indefinitely on individual enthusiasm. Academic incentives will eventually need to recognise forms of contribution that matter more in an AI-mediated knowledge economy. Publication, research quality and academic teaching will remain important, but institutions should also learn to recognise disciplinary currency, meaningful AI integration, high-quality external engagement, translation of research into application, student enrichment through external partnerships, creation of real-world datasets and research environments, and interdisciplinary collaboration. 

This does not mean reducing academic evaluation to a corporate performance dashboard — quite the opposite. It means recognising that scholarship itself is changing. A professor who brings contemporary industrial problems into a classroom, develops an AI-enabled research method, creates a field dataset, collaborates with a public institution, and turns that experience into stronger student learning may be performing an extraordinarily valuable academic function, even where none of it fits neatly into an old publication-counting framework.

2.8 The AI-era university must be more permeable

What results is a university that looks different from the traditional institution without becoming less academic. Not one where industry dictates scholarship, or AI dictates curriculum, or every professor becomes a consultant, or every student becomes a startup founder — but one where the boundaries between knowledge and application become more permeable. 

The university receives knowledge from the world, tests it, produces new knowledge, educates students through it, and sends knowledge and capability back into the world - which then sends new problems back. That circulation is what gives the institution continuing relevance.

2.9 And this is where the debate becomes larger than AI

Once the university is understood as a knowledge-circulation institution, its purpose can no longer be reduced to graduate employment, research rankings, startup creation, online instruction, or even technological innovation on its own. It becomes part of a much larger national capability system: an industrial university circulating knowledge between research and production, a central university circulating it between scholarship and national missions, a state university between scholarship and state missions, a regional university between academic disciplines and the communities, economies, and ecologies of a particular region, a specialist institution pushing a field toward the frontier, etc. These are different institutional forms sharing one principle — the university has to be connected to something larger than itself.

That is a much larger role than the lecture university, and a much larger one than the startup university too.

2.10 From teaching institutions to capability institutions

The transformation India needs, then, is not from offline education to online education, or from universities to AI platforms. It is from isolated educational institutions to capability institutions embedded in the systems around them. 

AI makes this more urgent, because it exposes the weaknesses of the old model, but it also hands universities extraordinary new tools for reinventing themselves. The university can become more personalised without being less communal, more digital without being less physical, more automated without being less human, more connected to industry without being subordinate to it, more engaged with government without becoming an arm of it, more technologically sophisticated without abandoning the humanities, more economically relevant without reducing every discipline to a job description. 

The challenge is not to protect the university from AI. It is to use AI to rebuild the university around its highest-value functions.

Part III: Building the Missing Middle

If the university is to remain a serious institution in the AI era, India also has to stop asking it to solve a problem it cannot solve alone: the transition from education to employment. For years, the higher-education debate has been dominated by a deceptively simple question — are our graduates employable? The question is legitimate but incomplete. A university can provide disciplinary foundations, analytical ability, technical knowledge and intellectual maturity; it can expose students to contemporary technologies and real problems; it can integrate AI into education and connect faculty with industry and government. 

But it cannot reproduce the full experience of working inside a factory, laboratory, hospital, bank, software company, infrastructure project, government department or research organisation — nor should it try. The workplace has its own forms of knowledge. A computer-science graduate may understand algorithms without having managed a production system. An engineering graduate may understand thermodynamics without having dealt with an unexpected equipment failure at an operating plant. A biotechnology graduate may understand molecular biology without knowing how a regulated manufacturing facility actually runs. An economics graduate may understand financial theory without having dealt with the consequences of a transaction error inside a live institution. Experience is what converts knowledge into judgement, and historically the first job did much of that conversion — a graduate entering an organisation on relatively routine, supervised work, making mistakes, learning organisational processes, and gradually acquiring professional judgement. 

The first job was not merely an employment outcome. It was an experience-formation institution, and AI is beginning to disrupt precisely that mechanism.

3.1 The AI-era experience bottleneck

The first generation of automation displaced repetitive physical and clerical work. This generation reaches into cognitive work — routine coding, testing, documentation, data processing and basic analysis can increasingly be performed or accelerated by AI. For an individual company, that is an obvious productivity gain. For the workforce as a whole, it creates a paradox: the economy will still need people who can evaluate AI-generated outputs, identify plausible but incorrect results, understand operational context, supervise automated systems, diagnose unusual failures, make decisions under uncertainty, integrate technical and organisational considerations, and take responsibility for consequential decisions — and all of that requires experience. 

If the routine work through which inexperienced people traditionally gained that experience disappears, where do the experienced professionals of the future come from? That is the experience bottleneck, and the answer is neither to preserve obsolete jobs to keep graduates occupied, nor to pretend a university can simulate every workplace. The answer is to build a stronger institutional layer between education and employment. India needs a missing middle.

3.2 From employable graduates to trainable graduates

This calls for a subtle but important shift in how graduate capability is understood. A graduate does not have to be immediately productive on day one; what matters is whether they can enter a structured environment and acquire occupational capability quickly. The objective should shift from producing employable graduates to producing trainable ones — trainability meaning disciplinary foundations, analytical ability, communication, digital and AI capability, problem-solving, learning capacity, professional behaviour and the ability to work with others, with workplace formation adding operational knowledge, domain-specific process, organisational experience, professional judgement and responsibility on top. 

Universities educate. Workplaces train. The institutions between them make the transition possible.

3.3 The education–employment bridge

The basic architecture can be straightforward: university education, then internship, then graduation, then apprenticeship or traineeship, then employment. 

The internship's job is mainly exposure — letting students encounter real workplaces, projects, laboratories and professional environments before they finish their degrees, so they understand how knowledge operates in practice rather than becoming immediately productive. 

After graduation, apprenticeships and traineeships provide the deeper layer of workplace formation, under whatever label a sector prefers — apprentices in manufacturing, graduate trainee programs in technology, structured early-career programs in banking, industry fellowships in research-intensive organisations. The institutional purpose is the same throughout: turning an educated but inexperienced person into a capable professional.

The same logic should extend to doctoral graduates. A doctorate should not be treated as a qualification whose natural destination is academia. Research-intensive companies, public-sector enterprises, advanced technology organisations, and government research organisations need people who can translate advanced research into application. A structured post-doctoral industry-fellowship could be particularly valuable in advanced manufacturing, semiconductors, biotechnology, materials, AI, energy, pharmaceuticals, climate technology, and industrial research, strengthening the university-industry relationship without forcing universities themselves to become industrial training centres.

Rapidly changing industries raise a further complication: some capabilities shift too fast to justify rewriting an entire degree every few years, which is where micro-credentials belong — short, recognised modules in new software tools, AI systems, industrial technologies, regulatory requirements, data analysis, new production methods, and sector-specific practice, for which online platforms are genuinely useful. But the right model here is integration, not substitution. A micro-credential can add a capability; it cannot supply the intellectual foundations, institutional relationships, laboratories, peer community and supervised experience a university provides. The choice is not degree or online course — it can be degree, plus online course, plus internship, plus apprenticeship, plus continuing learning. That is what a mature education system should look like.

3.4 Talent Transition Infrastructure 

There is a further problem curriculum reform cannot touch: opportunity is not the same as access. A student might secure an internship in Bengaluru without being able to afford temporary accommodation there. A graduate might land an apprenticeship in Pune only to find relocation costs eat most of the stipend. A trainee might get an opportunity in Hyderabad without family, friends, or any network in the city. This is a physical infrastructure problem sitting inside human-capital mobility, and one response is what might be what I have called Talent Transition Centres — shared infrastructure for people who have been selected for an internship, apprenticeship, traineeship, or fellowship -  combining affordable accommodation, connectivity, mentorship, peer networks, industry interaction, supplementary learning, professional events, hackathons, and career support. It should not become a recruitment agency; employers keep selecting their own interns, apprentices, trainees, and fellows. The centre simply makes it easier for people to take up opportunities that would otherwise be out of reach.

This is where municipalities enter the picture — cities are not just where companies happen to operate, but the physical environments through which workers reach them, which gives municipal governments a legitimate role in the infrastructure around talent mobility: helping provide land, electricity, water, sewerage, roads, public transport, connectivity and links to transport nodes, with state governments complementing this by mapping regional industrial and talent trends and coordinating transition infrastructure across cities. India has spent decades treating roads and logistics as infrastructure for the movement of goods. It now needs to treat accommodation, connectivity and mobility as infrastructure for the movement of talent.

Individual employers should continue to decide how many people they need and whom to recruit, but not every problem should be solved company by company. Industry associations can provide common infrastructure around competency frameworks, portable credentials, mentorship networks, industry hackathons, talent forecasting, micro-credential recommendations, and shared transition infrastructure - coordinating across sectors and cities in ways no single firm can. This matters especially for MSMEs — a large corporation can run a sophisticated graduate-training program, a small manufacturer cannot - and an ecosystem-level institution can let smaller firms participate in structured talent formation without each one building an expensive apparatus of its own.

The physical infrastructure will stay fragmented without a common digital layer, which is where a National Talent Mobility Platform could connect students, universities, employers, apprentices, trainees, fellows, Talent Transition Centres, industry associations, state and municipal governments, mentors, and micro-credential providers — not as another job portal, but as the connective layer between opportunity, mobility, accommodation, mentorship, learning and professional networking. A student with an internship in another city could find appropriate TTC accommodation; an apprentice could find courses recommended by the relevant industry association; a fellow could identify mentors in the relevant research ecosystem; a university could see broad patterns in where its graduates are entering work.

3.5 India also needs talent-demand intelligence

One more piece is missing from most education policy: a better sense of the broad professional capabilities India is likely to need over the next one to four years. This should not become a crude manpower-planning exercise — no institution can predict precisely how many people India will need in every profession four years out - given how uncertain technological change, investment cycles, global markets and entrepreneurial activity all are. 

The aim should instead be indicative talent intelligence: a rolling national outlook on which sectors are expanding, which professions are likely to grow, which qualification levels will be needed, where demand is concentrated geographically, which capabilities are emerging, and which existing occupations are likely to contract or transform — across a one-year horizon for immediate pressures, a two-year horizon for apprenticeship and postgraduate planning, and a four-year horizon for program, faculty, and infrastructure decisions. 

This could be run by a representative academic institution or association — potentially the Association of Indian Universities — in consultation with disciplinary societies, professional associations, industry associations, startup associations, GCC representatives, large employers, major investment funds, and relevant ministries. Investment flows themselves are a useful forward-looking signal: when large amounts of capital commit to a new industrial ecosystem, talent requirements often become visible well before the facilities reach full production. 

The resulting intelligence should guide universities, not command them — they should keep academic autonomy while getting much better information about the economy they are preparing students to enter.

Put together, the architecture becomes a connected system: investment and industrial plans feeding talent-demand intelligence, feeding university program and capability planning, feeding education, feeding internship, feeding apprenticeship or traineeship, feeding employment, feeding employer feedback, feeding the next talent-demand outlook — with a parallel track for advanced researchers running from doctorate degree to industry fellowship to applied research to specialised employment. 

This is not a pile of disconnected schemes. It is an institutional ecosystem.

Part IV. Beyond Sanyal's Sensationism 

Sanjeev Sanyal is right about something fundamental: India's university system cannot continue unchanged. But that is where the argument should have begun, not where it ends. Leaping from the obsolescence of some lectures to the obsolescence of universities; from the weaknesses of some academic cultures to the uselessness of entire disciplines; from the success of an institution like IIT Madras to a universal prescription for thousands of universities; from the usefulness of online learning to the presumed redundancy of physical knowledge institutions; from the rise of AI to the conclusion that institutional education should be bypassed; and from graduate unemployment to the idea that universities should essentially become startup or skills factories — none of that is radical thinking. It is insufficient thinking, and it reflects a narrow view of the economy itself.

India's future will not be built by startups alone. It will be built by startups, established corporations, MSMEs, public enterprises, research institutions, governments and millions of workers interacting across increasingly complex technological systems. The large industrial enterprises currently investing enormous sums in manufacturing, energy, infrastructure, electronics, pharmaceuticals and materials will need enormous amounts of knowledge, research, training and technological capability — a university system that ignores them because they do not fit a startup narrative is failing to see the economy actually in front of it. 

Nor does the employment problem disappear because an AI application can teach a student something: learning a programming language online does not teach someone to operate a production system; learning financial theory from an AI tutor does not give someone experience of a live financial institution; learning engineering principles through a digital platform does not teach someone how a complex industrial organisation behaves when equipment fails, supply chains break, or safety decisions have to be made under pressure. Knowledge and experience are complementary. Education and workplace formation are complementary. Universities and employers are complementary. The task is to build the institutions that connect them.

The university is not obsolete. Our conception of it is. India should not defend the university by preserving its inherited weaknesses — it should not protect rote pedagogy, weak research, outdated curricula, ideological insularity or disconnected academic cultures simply because they happen to exist inside universities. 

But neither should it mistake those weaknesses for evidence that the institution has outlived its purpose, which is precisely the mistake the legitimacy arc this piece opened with keeps making: judging the whole institution by its worst-performing part, or its most-publicised podcast moment, rather than by what it is actually capable of when asked for more than it is currently asked to give.

The university can become something much larger. It can educate, research, preserve knowledge, generate new knowledge, work with industry, work with government, serve regions, support national missions, preserve heritage, support ecology, develop AI-augmented disciplines, provide long-horizon research capability, and circulate knowledge across organisations and generations. It can produce graduates who are not expected to know everything on the day they graduate, but who carry the foundations, curiosity, and adaptability to become highly capable professionals through structured experience. The central policy question, then, should not be how to replace universities. It should be how to make India's universities worth having — a far more demanding question, and one that deserves a far more ambitious answer.

India does not need fewer institutions of knowledge simply because knowledge has become abundant. It needs better institutions for turning abundant knowledge into human capability, technological capability, industrial capability, public capability and social capability. Artificial intelligence can help enormously. So can online learning, startups, industry and government. But none of these makes the others redundant — the task before India is to connect them, and the university can be one of the most important institutions through which that connection happens.

The university of the AI era does not need to be smaller, less physical, less academic or less human. It needs to be more connected, more capable, more differentiated, more technologically augmented, and more deeply embedded in national life. The real reform is not the abandonment of the university. It is its transformation — from a relatively insulated place where knowledge is transmitted, into a living institution through which knowledge is continuously produced, tested, preserved, applied and circulated. That is not the end of the university, nor is it a legitimacy already spent. It is the beginning of a much bigger one.