Changing Our Paradigm

A change of paradigm is usually associated with a change in language. Take an example from my own field. Centuries ago, people spoke about insulators and metals to measure how they conducted electricity. Later, we introduced concepts such as band structure, band gaps, and the Fermi level. Today, we talk about the topology of bands. It doesn’t mean we were wrong before. It means we constantly develop better and better language to describe nature. That is how we enhance our understanding and create predictive power in science.

This is what we call the origin of complexity. The functionalities we require from technology are associated with a certain level of complexity. But in most current technologies, functionalities are situated on only one level. Take any functional system — an electronic component, a microprocessor, a robotic system, even a car. Remove one component, and it immediately stops functioning. The complexity, the functionality appears on only one level.

In biological systems, complexity and functionality are distributed across many levels. Biological molecules such as proteins have functionality. Cell membranes have functionality. Cells themselves, organs — each level contributes. This distribution allows to achieve far more complex behavior.

That is what we want to learn from nature. We want to delegate certain functions to the material level, so materials could generate energy themselves, perform certain tasks independently, perhaps even make decisions depending on the environment. We are not there yet. But many researchers are working on functional materials today.

At the moment, we do not distribute functionality across multiple levels in this way. For me, this question — how to achieve this — is connected to the origin of complexity, and even to the origin of life. We are still very far from providing a decent answer. We may think we understand how our bodies work, and people speak of neuromorphic computing, mimicking the brain. But in reality, we are still far from understanding biological complexity. The term “neuromorphic” can be misleading. We do work on neuromorphic computers here in this Institute, but they are still far from gaining the same level of functionalities and complexity as the biological systems.

That shift in perception was extremely important. We figured it out that monolayer materials — two-dimensional, one-atom-thick materials — can exist, be stable, and be studied. And they exhibit unusual quantum mechanical properties.

This is how scientific discovery works: you create a new paradigm, one paradigm is replaced by another. The paradigm that two-dimensional materials don’t exist, that you can’t arrange atoms in two dimensional crystals was replaced by the realization that they can exist — and that we can work with them.

Andre asked whether we could make a transistor out of graphite. It was obvious that we needed to take a very thin layer of graphite and then try to make a transistor out of it. We tried many approaches, none worked, and we pretty much forgot about it. One day, we were preparing samples for our scanning tunneling microscope. We used Scotch tape to peel the top layer of a graphite crystal — a routine cleaning method. Normally, you throw that away and use only the freshly cleaved crystal. But because we were thinking about this idea of how to make thin graphitic films, we thought: why can’t we try and make a very thin layer with this scotch tape? And the very first device worked! It worked very badly but it worked. That told us we were on the right path. It took another year before we produced our first graphene devices.

Of course, graphene gave rise to many applications. It has found applications in batteries, composite materials and electronics everywhere. But the fact that graphene changed our paradigm, that two dimensional materials can exist, is the most influential consequence of, of that work.

We do not work only with 2D materials, but they introduced a powerful concept: artificial materials, materials on demand. If you have access to these two dimensional crystals which are only one atom thick, you can assemble them with atomic precision into a stack and put one layer of one material, another layer of another material, then another one… And then you basically create a three-dimensional material which didn't exist in nature before.

By combining many different two-dimensional crystals, you can design entirely new artificial materials.

Some of these materials can change their behavior depending on their environment. This opens the possibility of creating active materials — materials that communicate with the environment. And the ultimate goal is really the functional materials, which can change their properties, which can react to the external stimuli and maybe just behave as biological materials.

I am very positive that it’s going to work and AI will become an important tool for the discovery of complex, dynamic materials that can evolve by themselves. We are not there yet. But I still believe that the role of a scientist will be very important. AI is a tool for us. And it's going to be a very powerful tool. But the role of the scientist, of the researcher will still be dominant.

Science really works by accumulating broad knowledge across many areas. These areas strongly cross-pollinate and influence one another.

Science must be developed as one, not as individual silos.

Editor’s note: This article has been faithfully transcribed from the original interview filmed with the author, and carefully edited and proofread. Edit date: 2026