Dr. Viktor Dörfler is a scientist, lecturer, consultant and philosopher who has spent the last 25 years as a scientist studying the great masters, including Nobel laureates and excellent chefs. (The term "grandmaster" is borrowed from chess to describe the highest level of knowledge in any field.)
Talk title: Grandmasters and Apprentices in the Age of AI
Every major technological advancement was expected to change education as we know it. Yet, if my grandfather came back, right now and here in this room, he would recognize the teaching in the classroom. Sure, it is not chalk on blackboard anymore, now it is PowerPoint or Prezi, but the essence is the same: someone who supposedly knows stands in front of an audience and preaches.
The biggest change that happened in teaching-learning was enabled by the internet and enforced by Covid: lectures took place online. But, hey, I said it was my grandfather, he had a TV and he was a clever guy, it would take him 2 mins to figure it out. And, you know what? This change is not really fundamental. It is still someone speaking to an audience…
But now AI is certainly changing the story, isn’t it?
Many expect AI, the “smart technology”, to do away with the teachers, it is argued that knowledge is “democratized”, it is there for anyone who wants it. Is this true?
Not quite. Information is what is available to everyone. Information is shallow knowledge. With reference to shallow knowledge Arthur C Clarke famously said ages ago that: Any teacher that can be replaced by a machine, should be. I agree.
Knowledge, however, can also be deep. And I am here with you today to explore what cannot be replaced and who cannot be replaced. What I want to talk to you about is an ancient, possibly the oldest form of teaching and learning, that I call the grandmaster-apprentice relationship.
There are a few reasons why I can tell you about the grandmaster-apprentice relationship. First, I have experienced it myself as an apprentice, and later I had my apprentices too. When I joined the workshop of my grandmaster, I became part of a team supporting C-level decision makers with AI. Soon, building on my multidisciplinary background, I acquired two distinct roles that I have continued to pursue for the last quarter of a century.
As a practitioner I was developing AI software and implementing AI solutions. In total we had about 160 AI implementations, and I participated in about 50 of these. Of course, this was all before the time of ChatGPT, it was what we call symbolic AI.
During the same 25 years, as a scholar, I was trying to understand how people think, learn, and use their knowledge. As part of this inquiry into knowledge, I have developed a conceptual model of the levels of mastery. As that model was purely speculative, I was told that I should conduct an empirical study, which I wanted to aim at the highest level of mastery, which was, at least to me, the most exciting.
Well, one way to do it would require about 200 million dollars, staff of about 50 people, and guys like Gary Kasparov at my disposal for about 6 months at a time… Perhaps you have already guessed that I had to find an alternative route…
What happened is that I gave a talk at a conference in Tokyo. I had no idea that the guy asking me a question after my talk was a Nobel-Prize-winning physicist. I gave him a sharp-and-short answer, everyone was laughing, including him. He approached me in the break – and by that time I was well aware that he was the physics Nobel Laureate, Martin Perl. He said: “I’ve liked very much how you talked, but I have no idea what you were talking about.” I replied: “It’s because the talk was not designed for you, if you give me 3 minutes, I’ll explain it the way that you get it”. Well, he was an experimental physicist, so he looked at his watch: “You have three minutes…”
I must have done something right, as we became friends. And this made me think: what if I assumed that Nobel Laureates are at the highest level of mastery… To say simply, they are as close as you can get to finding a well-defined group of grandmasters… So, I decided to embark on a journey of interviewing Nobel Laureates! I will not even tell you the story of how difficult it is to get funding for such a project – but in the end, a friend managed to get some money and we got started.
I was interested in how these grandmasters of science think and work and how they became what they are. Of course, you cannot just ask them how they think, as nobody can answer such a question. So, I had to be able to engage in an intelligent conversation with them about their research topic. You cannot imagine how much physics and economics and chemistry and biology I have learned… I have to admit that it was not easy, but it was also a lot of fun – imagine, I was getting paid to have conversations with quite a few of the smartest people of the world… What I am going to tell you about today, is what I have learned from these conversations about the grandmaster-apprentice relationships.
But why is the grandmaster-apprentice relationship so important in the first place? Well, when I started my interviews, I more or less thought that nobody can achieve the grandmaster level of mastery without going through a grandmaster-apprentice relationship, with the exception of a genius. I was wrong. Even the genius has to go through a grandmaster-apprentice relationship, although in such case, this may take a more subtle and symbolic form. I have realized this from my interview with Yoichiro Nambu, who was never allowed to actually speak to his grandmaster, due to class differences, even though he was regularly in his physical proximity. Then I have found confirmation in the literature, where Howard Gardner writes about Mozart who was arguably always better than his mentor Johann Christian Bach, yet Mozart went through the process. In short, it seems that there is no way around it, there is no shortcut: the only way to the highest level of mastery is through the grandmaster apprentice relationship.
However, the grandmaster-apprentice relationship is not the first step on the road to becoming a grandmaster, it is the last one. First, the future apprentice is inspired by an inspirational teacher. If you thought that grandmasters are rare, inspirational teachers are much rarer. From what I gather, a whole generation of physicists, not only the Nobel Prize winning ones, were inspired by one of only two people: Enrico Fermi and Richard Feynman. The inspirational teacher is certainly a grandmaster, but they may or may not be suitable for a grandmaster-apprentice relationship. For instance, Enrico Fermi was also great at bringing up apprentices, I was fortunate to talk to one of them, James Cronin. But Richard Feynman, for instance, never had any apprentices, as according to one of my interviewees, Roy Glauber, he could only teach anyone how to be Feynman, and nobody could be Feynman, as he was too brilliant. The inspirational teacher here does not inspire you to be a physicist, but to be a scientist. When chemistry Nobel Laureate Sherwood Rowland was a freshman, he asked senior students whose classes to take, and they told him “Whatever Fermi is teaching.” – “But Fermi is a physicist, I am a chemist!” – “So what?”
I don’t think I need to justify that the grandmaster-apprentice relationship works, if all the Nobel Laureates had such experience. Michael Polányi, from whom most of what we know about knowledge originates, suggests that this is the only way of transferring tacit knowledge.
So, okay, it is the way to become a grandmaster – but do we really need grandmasters?
Yes, I believe we do.
Grandmasters develop an immensely deep knowledge in their respective domains, and create new knowledge that may change the course of history. Let me try to depict, with broad brush strokes, how the grandmaster-apprentice relationship works, in order to help understand better what it is.
It is easy to understand why the grandmaster-apprentice form of learning is not so popular today; it goes against the idea that we are all equal. The god-like figure of the grandmaster imposes upon the apprentice, who, at the start of the process, knows significantly less than the grandmaster. You have to understand that this asymmetry is both unavoidable and necessary, and none of it implies that the grandmaster is more important than the apprentice. Not only because the apprentice will one day, if it all works out, become a new grandmaster, but the grandmaster learns in this relationship as much as the apprentice, often more.
If you enter a grandmaster-apprentice relationship, you need to accept this asymmetry. You also need to know that this relationship is exceptionally deep and personal, as the grandmaster needs to know what you think, try, want, even before you do, in order to provide that special personalized teaching and support, that is just for you, just here, and just now, and it is exactly what you need.
A large part of the grandmaster-apprentice relationship happens through observation. And this is a particularly difficult point. You observe your grandmaster, and see the greatness, how the grandmaster can stand on a personal ground, apparently independent from anyone else. You aspire to that greatness, and you want to copy it, you want to become like your grandmaster. However, if you imitate your grandmaster, you can only become a pale copy, and never a grandmaster on your own right. Therefore, it is a mistake to follow the grandmaster’s way. Of course, it is also a mistake not to follow the grandmaster’s way, since it is the grandmaster who knows… This tension leads to a struggle to follow or not follow the grandmaster’s way, and from this struggle the new grandmaster emerges. This is much easier said than done, it is a real struggle, and it takes a long time. But, it works.
When I started my journey of interviewing the Nobel Laureates, I have thought that there was only one kind of grandmaster-apprentice relationship; this is what I today call the traditional model. There is one grandmaster and one apprentice, or perhaps a few apprentices at different stages of development. This model might be the dominant one, many of my interviewees had this experience and the realization of how it worked.
The second form of grandmaster apprentice relationship we discovered together with Marc Stierand, who was back then my doctoral student, today a notable creativity scholar. He studied top chefs – actually we still study top chefs together. In haute cuisine, the dominant form is the “journeymen” or, as I like to call them, the wandering apprentices, who move from grandmaster to grandmaster and learn different things in different kitchens. This form is also not uncommon in science, although much rarer than the traditional one.
Roy Glauber has asked me during his interview: “Which university is better Oxford, or Cambridge?” I said I could not tell, and he suggested that no sensible person could. However, the science Nobel Prizes are typically coming from Cambridge. Why? Because there is a laboratory, a hot spot, there is something in the walls. Like this one at Stanford, where a Nobel Prize winning biochemist and his dean are serving ice cream to the students. It is like a community version of the grandmaster-apprentice relationship, there are several grandmasters and a bunch of apprentices, with a web of interactions between them. It sometimes switches into the traditional model, sometimes there are wandering apprentices, but mostly it works like a super-performing community of practice.
Finally, there is a very rare form of grandmaster apprentice relationship that I first have come across when talking to Daniel Kahneman, when I asked him about any mentors or grandmasters, he said: there was only Amos. The very next day I went to Princeton, talked to Eric Wieschaus, who had a similar experience with Yani. Two talented people, already at an advanced level of mastery, get together and somehow become each other’s grandmasters and apprentices. Eric and I coined the term together right there during the interview: mutual apprenticing. I am not sure that there are no further variants of the grandmaster apprentice relationship, but these are what I have seen.
As you now clearly see, from the first Arthur C. Clarke quote on, I am arguing that only shallow teaching and only shallow teachers can be replaced by AI. That is why I am extremely passionate about the grandmaster apprentice relationship, which is, as far as I can tell, the only way to the highest level of mastery. AI can tell you the statistical frequency of what is. The grandmaster tells you – what can be.
Thank you.
Professor of AI Strategy