AI and New Ways of Knowledge Transmission

At India-Spain Conference on Higher Education.

19-20 February 2026, New Delhi, India

It is a privilege to take part in this discussion at a moment when artificial intelligence is not simply advancing — it is accelerating at a pace that feels almost exponential.

I am perfectly aware of the leadership of Indian institutions in the development of AI, and of the great potential of the Indian academia and industry to continue and reinforce this leadership for different reasons: (1) the background and excellence of the education and preparation of tens of thousands of researchers, engineers and professionals in universities, research centres and in the industry, and (2) the great efforts done by public bodies that have resulted in the existence of great analysts working for private and public bodies, and very rich datasets that will result in further development of important applications, and in the progress of the AI tools themselves. 

So it is a great honour to be here participating in this panel with our great colleagues from the Indian academia, and from Spanish universities. 

We are often told that AI will transform education, research, and industry. But today I would like to focus on something more fundamental: AI is transforming the way knowledge itself is produced, validated, and transmitted. 

And that transformation directly affects the role of the university. Today, I am going to address some of the basic points that, in my opinion, must be taken into account, and that in different ways are already in the road map of UPV, and of many other Spanish academic institutions.

Let me begin with research. Artificial intelligence does not merely accelerate scientific work — it changes its structure. 

AI can conduct literature reviews in minutes, generate hypotheses, design simulations, analyse vast datasets, or assist in drafting scientific papers, but not only ...

We might say that the research cycle is compressed, but the truth is that something even deeper is happening. The bottleneck of science is shifting: for decades, computation and data processing were constraints. Today, with AI, those constraints are diminishing. 

The new bottlenecks I foresee are, at least, the formulation of meaningful questions, the access to high-quality data, the always pending or not enough developed Interdisciplinary integration, and the "conceptual imagination". Because AI is shifting the bottleneck of science from computation to creativity and imagination.

This means that research excellence will increasingly depend not on who can compute faster, but on who can define better problems and integrate knowledge across domains.

AI does not replace research. It raises the threshold of excellence. 

Traditionally, transfer followed a linear path: Research . Patent. License. Product. But AI disrupts this sequence and creates different paths like: Data. Model. Platform. Continuous iteration, with retro feeds and amplifies the outputs transferred. 

AI can clearly compresses the distance between laboratory and market, and thus AI allows, for instance, that prototypes are simulated before being built, the use of predictive models to reduce technological risk, the development of digital twins or algorithms that can be transferable assets. This means that innovation cycles shorten dramatically and that time-to-impact is shrinking. 

But these facts also raise new institutional and very important questions like: Who owns AI-assisted inventions? How do we protect algorithmic intellectual property? And how do we regulate co-created outputs between humans and machines?

Our current Tech Transfer offices must evolve. There is the need to understand models, data governance, algorithmic licensing — not only physical technologies. AI is redefining what is transferable.

This new situation must be seen as a scenario that opens new opportunities that we must integrate in our strategy. And there are three major opportunities. The first would be to gain capabilities as AI Augmentation Hubs making research centres in our universities to become platforms that provide advanced AI capabilities to industry, domain-specific model training, validation and certification services and that made up sectorial data ecosystems. Thus, the research centre is not just a generator of knowledge. It becomes an amplifier of intelligence across the economy where collaboration with the industry will be fundamental. 

The second has to do with the view of Data as a strategic infrastructure. We know that high-quality, curated, ethically governed datasets will become strategic assets. And our research centres that structure and govern data responsibly, something that will be indispensable. 

And there is a third opportunity that might be seen also as a need. Our labs can become "human–machine Laboratories" a new kind of hybrid labs where co-work human researchers and autonomous AI agents, and where continuous simulation environments work in parallel. 

This model could multiply research productivity while reducing cost and time. But we must acknowledge a risk for our institutions when no enough computational infrastructures are available. In this case AI might introduce also a new form of inequality between universities with digital capacity and those without it. This is not only technological — it is geopolitical. Alliances among institutions and universities - internationally - might help to avoid such undesirable scenario.

I will finish with some concluding remarks that I consider of paramount importance. We must address the uncomfortable but unavoidable question that AI is increasingly capable of automating many current tasks like legal drafting, financial analysis, coding tasks, administrative reporting or standard diagnostics, and this completely restructures intellectual labor. Thus, in this AI era, productivity depends on how effectively one amplifies intelligence with intelligent systems. We must face that graduates will not compete against AI, instead they will compete against people who know how to use AI better. We, universities must redesign curricula to allow our graduates facing new scenarios we do not foresee and technologies yet to come. 

And of course this raises another question: Can AI replace researchers? Of course AI can replace many tasks that researchers performed like routine data processing, incremental literature synthesis, standardized technical writing and low-level replication research. But it cannot easily replace others like conceptual breakthroughs, cross-disciplinary synthesis, ethical judgment or the necessary strategic scientific leadership.

Thus, I believe that something profound may occur: an average researcher, empowered by AI, may achieve levels of productivity previously reserved for top performers.

And finally, how should Universities prepare for this scenario?

We have to be ready to assume fewer purely routine research roles, put greater emphasis on originality, increase our interdisciplinary capacity, and higher standards of conceptual innovation. And to do that we must:Prepare our infrastructure investing in high-performance computing, in secure AI environments, in data governance systems and in international digital research platforms. 

On the other hand, we need a real evaluation reform of the performance of our researchers. AI increases the output volume so we must evaluate impact, integration, transfer, and responsible innovation, not merely publication count.

Additionally doctoral education must include AI-assisted methodologies, human–machine collaboration skills, responsible AI use, and cross-domain literacy.

And all of this must be developed in a strategy of universal AI literacy that reaches everyone in the university community.

Thank you.