Michał Karbowiak: Do you know how to spot a genius?
Prof. Leszek Pacholski: I think it's incredibly difficult. The closest I ever came to true genius was many years ago in Paris, at a lunch hosted by Stanisław Ulam (laughs). But by then, he was already a world-renowned scholar, so he didn't exactly need "spotting."
Did you feel intimidated meeting such a brilliant mathematician - a veteran of the Manhattan Project, co-creator of the hydrogen bomb, and someone whose work now underpins the development of artificial intelligence?
I was highly stressed, but not because of Ulam's genius.
Then why?
When my Parisian friends found out I was meeting Ulam, they insisted on organizing and funding the lunch. Among them was Roger Godement, an outspoken pacifist.
How did it turn out?
It was remarkably pleasant and civilized. Ulam and Godement got along famously, and by the end of the lunch, they were exchanging phone numbers (laughs).
Ulam and his brilliant peers from the Lwów School of Mathematics were legendary for solving complex mathematical problems on restaurant napkins. Did that happen?
Not this time; we managed without it (laughs). The conversation was fairly standard, but the memory of that meeting has stayed with me to this day.
The Summer Assignment
Do you remember any student of whom you thought: This is a brilliant mind, a true genius?
Every now and then, you encounter students who stand out for their distinct way of thinking. In fact, you can usually spot them as early as their freshman year.
During his freshman year, you taught Aleksander Mądry, who is now a professor at MIT and one of the core scientists at OpenAI.
I remember that after one of my courses, I gave Olek (the Polish diminutive for Aleksander) a summer assignment. He was only a first-year student at the time. The assignment was an open problem in database theory from a yet-unpublished paper by two Silicon Valley researchers. Right after the summer break, Olek came back with a solution so well-written that, with virtually no revisions, it was ready for publication in Information Processing Letters, a reputable Elsevier journal.
Did you know right then that he was exceptional?
It was obvious he was immensely talented, hardworking, and possessed vast intellectual potential. But above all, he wanted to build something that mattered. It's no surprise he chose a career in academia and theoretical computer science. He grew as a researcher in an institution where this field was incredibly strong, surrounded by highly influential professors. Looking back, however, I realize I was inadvertently misleading my students - Olek included.
In what way?
I didn't fully realize it at the time, but I wasn't a particularly good lecturer. I was chaotic and often lacked a structured syllabus; I could easily lose even the bright students. However, alongside standard courses, I would teach a seminar every year on whatever I or my colleagues around the world were currently researching. These classes drew highly ambitious, often very young students. The lectures themselves weren't polished, but I was deeply fascinated by the material, and that raw energy was infectious. The "deception" - entirely unconscious on my part - lay in convincing them that what we were discussing was the most vital research on earth, pursued by the world's most distinguished minds.
Wasn't it?
To some extent, yes. The researchers whose work I was presenting were based at top-tier institutions, like UC Berkeley or MIT. I had spoken at Berkeley about my own findings, and graduate students at MIT were reading them. But the work wasn't quite as monumental as I believed then, or as I led my students to believe. Even so, I managed to convince them that we were equal partners with the finest research centers in the world.
Did it work?
Years later, Aleksander Mądry told me that this atmosphere was precisely why he loved coming to my classes. So I suppose it did (laughs).
Did you encourage him to pursue his PhD at MIT?
He applied only to MIT and Stanford, and I fully backed his decision. When his first application was rejected, he spent a year working tirelessly to ensure he wouldn't fail again. He could have easily coasted into any other excellent university, but he was determined to settle for nothing less than the best.
Weren't you tempted to keep him at his alma mater? Perhaps he could be doing great things here in Poland rather than for MIT and OpenAI.
Genius never flourishes in isolation; it requires the right people and the right environment. This is absolutely critical. There are places where knowledge accumulates within the very walls - universities, research centers, and companies with a dense concentration of talent in a specific field. If you want to grow, you need to be there. Otherwise, catching up with those who have that privilege is nearly impossible. You cannot do ground-breaking work in a field like artificial intelligence on your own, cut off on the periphery. It's a non-starter, and these brilliant minds know it.
This isn't just true for the University of Wrocław; it applies to far more renowned institutions. Do you know what Łukasz Kaiser told me after defending his PhD at RWTH Aachen?
What did he say?
To paraphrase: "They want me to stay in Aachen for a few more years. But there is nothing more they can teach me."
The Transformer Architect
Łukasz Kaiser is one of the few protégés you praise publicly. While at Google Brain, he co-authored the seminal paper "Attention Is All You Need", which has over 240,000 citations (as of May 2026) and serves as the bedrock of modern AI and large language models. He won the NEC C&C Prize - often dubbed the 'Japanese Nobel' - and Nvidia CEO Jensen Huang called him a "hero." Kaiser is widely regarded as a pivotal figure at OpenAI.
Can you believe that a man with those achievements didn't even make it into the elite mathematics track at the prestigious High School No. XIV in Wrocław? (smiles). He graduated from the school, but from the English-language track. Above all, Łukasz is an incredibly warm, approachable person - the exact opposite of the stereotypical, detached nerd.
I remember many years ago, a professor of Polish linguistics at the University of Wrocław asked if I could recommend a student to help her build and scale a digital humanities research tool. Łukasz agreed to help. She was completely enchanted by him. She was delighted that someone with such profound knowledge could be so profoundly "normal."
When Łukasz was writing his master's thesis under your supervision - Confluence of Right Ground Term Rewriting Systems is Decidable - did you anticipate this kind of monumental trajectory in science and tech?
The work he did back then was highly niche, abstract, and difficult. Of course I couldn't predict the career he would have. Back then, no one could have foreseen the explosive evolution of computer science we are witnessing today. I do, however, vividly recall a lecture Łukasz gave a few years after he started at Google.
It came just a year or two after a presentation on neural networks at our university by an American professor who was two generations older than Łukasz. That older professor's talk left me convinced that the field was dying out. But after listening to Łukasz, I had to completely re-evaluate. It became glaringly obvious that the field was on the cusp of an explosion. And today, we have an explosion beyond anyone's wildest imagination.
Yet, you and Kaiser have a fundamental disagreement. In an article for Wszystko co najważniejsze, you wrote a piece titled: Don't worry, ChatGPT won't revolutionize science. Kaiser, meanwhile, believes AI will completely transform it.
There is no disagreement anymore. That revolution is already happening.
So you were wrong?
I shouldn't have written that article. It was far too early to offer a definitive critique, and I wrote my opinion in haste.
How do you view this revolution now?
AI can drive immense progress, but it carries profound risks that we must acknowledge. I recently read a discussion about Anthropic's unreleased Claude Mythos model. Apparently, it is so powerful it can uncover critical vulnerabilities in IT systems in seconds. In the wrong hands, that's catastrophic. Yet, experts note that cyber vulnerabilities are still a lesser threat than the potential for AI to design and engineer a deadly pathogen. Software patches can be deployed relatively quickly. But neutralizing a new virus? How long does it take to develop a cure?
So you are a deep pessimist when it comes to AI?
No, I am simply highlighting both the immense possibilities and the systemic threats. For instance, we need a much more serious conversation about the societal impacts of automation. AI won't replace bricklayers, but what about the entire white-collar sector? There is a severe lack of reflection on this.
At the same time, I realize that my former students are sitting very close to the center of global AI decision-making. Aleksander Mądry was specifically tasked with forecasting the negative fallout of AI and assessing which economic sectors will automate first. Łukasz Kaiser knows better and faster than almost anyone where ChatGPT is heading technically.
The processes we are talking about have transcended research and business - they have massive societal and geopolitical consequences. Furthermore, this isn't happening in a vacuum. You have OpenAI, Google DeepMind, Anthropic, xAI, rapid developments in China, and emerging efforts in Europe. Given this landscape, the wisdom and ethical responsibility of these tech leaders are paramount.
They wield immense influence, advise world leaders, and shape the global economy. The Economist once surveyed its readers on which tech titan posed the greatest threat to humanity; Elon Musk won by a landslide, while Demis Hassabis of DeepMind was viewed as the most measured and balanced. Personally, I would much prefer it if the measured individuals were steering the course of AI in the coming years.
Inside the AI Empire
It's up for debate whether that balanced approach exists within OpenAI itself. Investigative journalist Karen Hao, author of Empire of AI: Dreams and Nightmares in Sam Altman's OpenAI, suggested in a recent interview that the drive to build Artificial General Intelligence (AGI) at OpenAI has become almost theological - that top engineers genuinely believe they are building a "machine God."
I haven't read the book or seen the interview, so I can't speak to what her sources told her. I can only tell you that I have never heard anything remotely like that from either Aleksander or Łukasz.
They are incredibly grounded, emotionally stable scientists with proven track records. For them, OpenAI is simply a dynamic environment where they can push the boundaries of a field they have dedicated their lives to. I think Aleksander put it best when he told me he was joining OpenAI simply because "that's where the action is."
Hao also argues that OpenAI was never truly meant to operate as a non-profit. She claims Sam Altman and Elon Musk strategically used the "humanitarian mission" narrative from the start to differentiate themselves from a commercial behemoth like Google, attract top-tier talent, and ultimately pivot to a massive for-profit model anyway. Have you ever raised these doubts with your former students?
No, I haven't. Our interactions in recent years have been strictly professional and operational. I can't imagine bringing up corporate ethics during a brief phone call or video sync. Perhaps it's a conversation for when we eventually grab a beer together, either back here in Poland or out in San Francisco. I've been planning a trip there for a few years now.
I should also note that the "commercial Google" she criticizes heavily funds foundational research and open-source projects. Many core breakthroughs that OpenAI relies on today were actually invented at Google. If I recall correctly, it was the agile, startup-like energy of OpenAI that drew Łukasz in, far more than any grand "mission."
Do you feel you played a defining role in shaping Aleksander Mądry or Łukasz Kaiser?
I systematically overhauled the Institute of Computer Science at the University of Wrocław, transforming it into an exceptional hub of learning, even by global standards. I have no doubt that the rigorous, inspiring atmosphere we built there shaped them. To what extent? That's for them to judge.
Every year, phenomenal talent graduates from our institute. Once, quite unexpectedly, I received a rather dramatic tribute for introducing an exceptional researcher to the global community.
For whom?
Michał Moskal, who is also at OpenAI now after spending many years at Microsoft Research. I ran into one of his senior colleagues at an international conference, and the man literally fell to his knees to thank me for mentoring Michał. It was a joke, of course (laughs), but telling.
According to Wikipedia, you and Michał Moskal actually co-created the programming language Nemerle.
I have no idea who wrote that entry. Michał designed and wrote Nemerle entirely on his own while he was still a student. It was a brilliant intellectual exercise, a language built on elegant type theory. It wasn't designed for mainstream corporate programming; like many languages created by academic researchers, it requires a level of abstract thinking that is quite rare in everyday software development.
Missing the Intellectual Tension
Did you ever try to convince Michał Moskal to return to Poland? Perhaps by finding a wealthy benefactor or sponsor?
I asked him bluntly once. He told me money wasn't the issue at all - he was already making a very comfortable salary at the time. He simply said, "I don't have anyone to talk to over there."
Meaning what?
I only truly understood what he meant a few years later when I had dinner with his colleagues. To put it metaphorically: while we in Poland are trying to figure out how to take the next small step, they are busy planning how to conquer the world.
Is Poland - as a medium-sized European nation - permanently relegated to merely exporting its greatest minds?
Brain drain isn't inherently bad. The real issue is that Polish academia remains isolated from the global circulation of talent. We should absolutely let minds like Mądry, Kaiser, or Moskal go out into the world, but we must also be capable of attracting global talent in return - not just returning Polish expats, but researchers of all nationalities and backgrounds. That is the only way to generate the vital intellectual friction, the cross-pollination of ideas and inspiration required for breakthrough science.
Why aren't we doing that?
That is a deeply systemic issue, perhaps best saved for another day. But in short: we rarely hire researchers from abroad, or even from other Polish institutions. At my own Institute of Computer Science, 80 percent of the faculty are, directly or indirectly, my former students. I'm not proud of that. I've tried to recruit from the outside, with very little success.
Salaries are a massive barrier; it is incredibly difficult to convince university bureaucrats to approve competitive packages for exceptional talent, even when the budget allows for it. The other major hurdle is the habilitation - a post-doctoral degree required in Poland but non-existent in the Anglo-American system. Without it, a scientist cannot formally advise PhD students or build an independent research group. International stars aren't going to accept being relegated to "assistant supervisors."
Outside the System
Is there no way to fix the system?
In 1972, as a 27-year-old researcher, I was invited to UC Berkeley. Even coming from a poor, communist bloc country, I didn't feel out of place. I met world-class scholars, saw how a thriving ecosystem operates, and couldn't comprehend why Polish universities couldn't replicate that model. Funding is only part of the equation. After I stepped down as rector in 2008, I poured immense energy into lobbying for higher education reform in Poland. Ultimately, I came to a sobering realization.
Which was?
If we want to build a truly world-class research university in Poland, we cannot do it within the constraints of the existing higher education framework. It has to be built completely outside the current system. To attract and retain talents like Kaiser, Mądry, and Moskal, we need world-class mentors who themselves were forged in the globe's finest institutions.
Geniuses, perhaps, like yourself? After all, you have been highly decorated for your scientific achievements.
My high school physics teacher, with whom I had a pretty good relationship, once told me: "Pacholski, you are no Einstein." And honestly, I've stuck to that ever since.
