“Good AI genes” – conversation with prof. Piotr Sankowski, director of the IDEAS Research Institute

Within a year, the IDEAS Research Institute grew to over 120 employees, 16 research groups, projects for the military and offices. How do you evaluate this start and what have you learned this year?

Assessing this start, I feel great satisfaction, especially with the pace of team building and how great our employees are. We are talking here about researchers who had every right and opportunity to pursue a career outside Poland, and yet they chose our center. For us, the ultimate test of quality is not only recruitment, but external, independent verification on the global scientific market. Regular publications by our teams at the most prestigious global AI conferences are hard proof that our work does not lag behind Western initiatives. I also realized that there is still a lot ahead of us, but our appetite definitely grows as we eat and carry out very interesting projects.

ELLIS is today the most prestigious AI league in Europe. Germany has Tübingen, Finland has Helsinki. What role can Warsaw play in this game for the future?

Poland educates great computer scientists, which is due to our outstanding mathematical traditions. Unfortunately, for years we have lacked an institutional anchor in the international AI environment. Such a pillar is joining and active participation in the prestigious ELLIS network, which we treat as one of the main strategic goals of IDEAS.

Establishing the first full-fledged unit in Poland in the form of the ELLIS Institute would be a strong and unambiguous signal to the whole world that we are doing great research in AI.

This would mark our final entry into the heart of the European AI research ecosystem. Establishing such a unit would certainly contribute significantly to reducing the phenomenon of brain drain.

You are the only Pole among the 40 experts appointed by the UN Secretary General to the global scientific panel on AI, selected from over 2,600 candidates from around the world. When you sit at this table, how is Poland perceived by the other members?

This is a great honor for me, but also a serious obligation as a researcher representing our region, helping to create independent assessments for the global panel established by the United Nations. When I sit at this table with representatives of great powers, I see that the perception of Poland is full of paradoxes. On the one hand, we are widely perceived as an incredibly strong pool of brilliant engineers and algorithmists. On the other hand, our dramatic lack of proportional state power is clearly visible – we lack large domestic models, huge investment capital and national infrastructure. We are a powerhouse when it comes to brainware, but we still lag behind when it comes to organization and scale of capital.

Good AI genes

The main theme of this year’s Wprost Innovators is “good AI genes”. You often talk about the need to build a Polish AI ecosystem. What are its strongest foundations today?

Our foundation is undoubtedly formed by outstanding people – Polish mathematicians and computer scientists who win international Olympiads and form the core of Big Tech companies. But the “genes” themselves will not survive in an unfavorable environment. The main challenge in Poland is to create an ecosystem in which the best scientists come to Warsaw or stay there out of their own choice, and do not leave due to lack of prospects. This absolutely requires stable, long-term financing. We must change the current model of science financing in Poland to encourage long-term, deep research necessary to create global innovations.

In Silicon Valley it is often said that technology is born where science, capital and the courage to take risks meet. There has never been a lack of courage in Poland. So why is it still so much more difficult for us to build our own technological champions?

There is no shortage of courage in Poland and never has been, we as a nation are able to approach the greatest intellectual challenges with bravado. However, the difficulty in building our unicorns lies in market barriers and extensive bureaucracy.

The domestic venture capital market still has problems with supporting capital-intensive, deep tech and lacks the scale to support large, risky projects.

In turn, foreign investors are faced with laws from a different era – they cannot understand that they are often forced to appear in person at a notary in order to invest millions in a modern company. These problems make our startups run away to look for capital in more friendly jurisdictions, and eventually they start operating in Silicon Valley.

Development and strategy

You often mention that there is still a “window of opportunity” for Poland in the field of AI. How long will it stay open?

This window of opportunity decreases with each quarter, leaving more and more specialized areas to be developed. Taking into account the pace of consolidation of the global market by large corporations and closing access to their models, we measure this key time to build our technological sovereignty in a maximum of 3-5 years. If we do not develop intellectual property and invest strategically within this time, we will be doomed to be dependent on licenses from outside.

The government wants to invest billions in AI gigafactories. You warn that without parallel investments in people, this money may work mainly for global players. Who should tell the government this directly and how?

This must be communicated to decision-makers with complete sincerity, and completely outside the current political dispute. The truth is that much of the European computing infrastructure already in practice performs operations for US companies.

If we build large server rooms but do not invest in experts who can use this power to create their own algorithms, we will in fact make a gift to foreign corporations.

As an expert community, we must loudly articulate to advisory bodies that bare walls and cables do not create innovation, but are created by talented people.

Politicians like numbers and count cable kilometers and teraflops. You count people. How to convince decision-makers, not only in Poland, to change this perspective?

The strongest argument that politicians around the world pay attention to today is absolute state security and technological sovereignty. I convince them to apply the iron rule of proportion: for every zloty from billion-dollar tenders spent on equipment (hardware), there must be exactly the same amount in long-term investment in human capital (brainware). Without such permanent financing of independent research units, there is no point in creating our own sovereign competences. Hardware becomes obsolete rapidly, while algorithms and human capital provide a lasting advantage.

What does this proportion look like in Poland today, how much do we spend on infrastructure and how much on people?

The comparison of these numbers is downright shocking and shows our gigantic, objective disproportion. While billions of dollars are publicly spent on building computing infrastructure and supercomputers, the budgets of key, market-leading entities dealing with “brainware” are appallingly small. For example, the budget of IDEAS is approximately PLN 40 million per year. Considering the infrastructure amounts, this is an almost symbolic amount. This image brutally makes us realize how much our system underestimates the work of human minds that underlie every modern technology.

AI on a full-time basis

IDEAS has implemented a system for analyzing notarial deeds, which shortens the work counted in hours to a few minutes. Is public administration ready for AI or does it still need to be convinced?

The situation is changing dynamically. Public administration and local governments no longer see artificial intelligence as a risky curiosity and begin to see it as a critical tool for operational efficiency.

The ExtrAct system, which we worked on in the context of extracting useful data and automating work with notarial deeds (crucial for tax registers, for example), is a great example of streamlining burdensome bureaucracy.

However, you need to be realistic – overcoming organizational barriers, building trust in new tools and integrating them with outdated office systems must take some time. However, our very advanced talks with local governments show that we are on the right track.

Energy sector, army or maybe administration? Where do you think AI will make the biggest revolution in Poland within 3-5 years?

In my opinion, the revolution will affect all of them simultaneously. In the energy and financial sectors, complex markets require precise modeling, and research on anomalies in large systems is carried out by our “Sigma” team led by Dr. Piotr Tempczyk. For the military, artificial intelligence is a condition of absolute operational survival – from reconnaissance analysis to drones and completely own autonomous systems. In administration, the above-mentioned document automation and improvement of existing interfaces have great potential.

What we need to understand as Europe

If you had an hour with European leaders tomorrow, without protocol and without cameras, what would you advise them?

I would tell them quite honestly and firmly that in-depth global analyzes expose our catastrophic fallacies. Europe has already slept through the previous Internet revolution and has not developed its own BigTechs. Now, unfortunately, the focus is on managing innovation through regulation rather than supporting it. At the same time, Silicon Valley is creating powerful new AI innovations and taking over our markets. I would advise radical cuts in polite small grants in favor of mass concentration of huge investment capital on leading European multi-year consortiums. I would also emphasize to them to stop being afraid. AI is not a job killer, it is a mechanism for driving progress, which, based on the market-based Jevons Paradox, will ultimately lead us to increased demand for highly skilled and better-paid developer skills.

What advice would you give to young Piotr Sankowski, who is wondering today whether it is worth building a career in AI in Poland and whether there is a place for him here?

I would give tough, realistic advice. I would not hide the fact that the path to creating a revolutionary deep tech project in the Polish administrative landscape is painful and requires a lot of fighting against unnecessary bureaucracy. But my main, authentic argument would be: by going to the West to work for a technology giant, you have a huge chance to become simply a well-paid, anonymous engineer implementing other people’s ready-made strategies. By deciding to stay in Poland, you participate in creating a modern economy absolutely from scratch. You gain an incredible sense of decision-making and can become a pioneer, a market leader capable of independently building a company that defines the rules of defense, science or medicine.

Wprost Innovators is an award for those who are transforming Poland. If you were to indicate one condition, one necessary, for Poland to be a real innovator in Europe in a decade, what would it be?

The final, marginal and only absolutely necessary condition would be the deep courage of those in power to think about the state in a fully long-term way. It must materialize in the enduring, iron-clad stability of inviolable financing of science underlying commercialization – a model known to us as Pasteur’s Quadrant.

And if I were to use a symbolic and tangible goal to achieve over the next few years, I would point to the establishment, anchoring and massive financing of the central Warsaw ELLIS Research Institute.

This would seal our path towards setting our own standards in the AI ​​market in a subjective way.

Interviewed by: Aleksander Wicka