Back
Wojciech Zaremba
Cofounder, OpenAI

Wojciech Zaremba – Twórca ChatGPT w Kluczborku | Spotkanie OpenAI & AI 1 sierpnia 2025 w hali OSiR

🎥 Aug 01, 2025 📺 KluczborkPL ⏱ 85m
Pełna hala OSiR w Kluczborku gościła Wojciecha Zarembę – współtwórcę ChatGPT i współzałożyciela OpenAI, urodzonego w ...
Watch on YouTube

About Wojciech Zaremba

Wojciech Zaremba, co-founder of OpenAI and head of language and code generation teams, discussed the iterative deployment of AI systems in a recent podcast appearance. He stated that deploying slightly better versions of models and allowing the public to probe them helps reveal fundamental issues. Zaremba also expressed concerns about artificial general intelligence (AGI) and advocated for this gradual approach to avoid unforeseen problems. Zaremba described three levers—compute, algorithms, and data—that he believes are multiplicative in building intelligent systems. He commented on capitalism’s ability to assign monetary value to activities but noted that the system lacks assigned costs for things like clean air or freedom from distraction, allowing corporations to "pollute them for free."

Source: AI-verified profile updated from Wojciech Zaremba's recent appearances. Browse all interviews →

Transcript (50 segments)
H
Host0:01
I would like to warmly welcome everyone to a special event, a meeting with Wojciech Zaręba, an artificial intelligence genius, co-founder of the Open AI company, who comes from Kluczbork. History has come full circle. Today Wojtek San Francisco returns to our family home. I warmly welcome you all. I welcome the invited guests, representatives of the district, municipal and provincial government authorities. I would like to extend a particularly warm welcome to the family of Wojtek Zaręba. Great people who support him every step of the way. I also warmly welcome his parents, who shaped his dreams here in Kluczorek. Hello to the friends and teachers I met along the way. And most importantly, I would like to warmly welcome Mr. Wojtek Zaręba and Dr. Maciej Kawecki, who will lead today's meeting.
Ladies and gentlemen, before we start talking, we have prepared a surprise for Mr. Wojtek and for you. I invite you to a unique screening of a film from Wojtek Zaręba's childhood. This is a unique material that will allow you to look at its beginnings from the perspective of family memories. It is a sentimental journey to the times when dreams were just being born.
N
Narrator2:41
Some kind of smoke powder was made and ignited. Wojtek invented a bomb and this bomb exploded in the middle school locker room. He soiled all the students' jackets in their cubicles. He had very kind ladies here in Luczbork at the pharmacy and at the Sanitary Inspectorate. And these experiments did not always end well. He dragged various things down to the basement for us and everything was fine until the explosion occurred. It's good that he had glasses, because he would have lost his sight. For me he is just little Wojtuś. Always beloved little Wojtuś. And when I come across an article from time to time, God, this is impossible, this is unbelievable. He was constantly searching for something. He had a question mark written all over his face. He kept asking. What he heard once, he remembered. He was able to put in a lot of work, from what I remember, he devoted 4 to 5 hours a day to mathematics, preparing for the Olympiad, he didn't have to get A's in everything. I never thought that Wojtek would be quoted more often today than the classics. When you talk to him and talk about numbers and programs, he says that he simply sees it.
I led the pedagogical council that sent Wojtek to Warsaw for gifted children. Wojtek returned from Belweder, where President Kwaśniewski awarded him a scholarship for talented children. For dear Mrs. Szot from Wojtek Zaręba. When it comes to mathematics, I quickly realized that his mind was far ahead of the program. Therefore, I decided that he would partially carry out the tasks independently. At that time, the first-grade class had four exercises for the entire year and he completed these exercises, i.e. the first-grade program, in the first half of the year. In the second half of the year, we had already completed the second-grade curriculum, but then I consulted with Wojtek's mother, the school principal, and the counseling center, and it indeed turned out that his mind was beyond that of a seven-year-old, and I received permission to conduct individual math classes, but in third grade. During breaks between classes, Wojtuś was very eager to approach me, talk, and tell me about his interests in computer science, mathematics, and especially chemistry, and he also conducts experiments in the basement of his house. Then, a little scared, I decided to put him in touch with our chemistry teacher. We met after classes. Well, the result was that, as a sixth-grade primary school student, he won the elimination round in Kruszbork and qualified for the chemistry olympiad for middle school students and in this olympiad he became a finalist. In middle school, these successes were annual, so the title of finalist in chemistry olympiads appeared. He came to me during a break and asked me if he would like to learn programming. During the next break we sat down at the computer and he actually wrote three lines of program text and the word Wojtek appeared on the screen. We took part in one of the student program olympiads. Wojtek was so surprised when he took second place, because his PowerPoint presentation won. Wojtek, I remember, started his presentation. I don't deal with [ __ ]. He started talking about fractals, he started talking about different kinds of loops. The result was that no questions were asked of him from the audience. Did Wojtek surprise me? Of course he surprised me, because some of the solutions, some of the algorithms he wrote were very innovative. They were constantly poking a stick into an anthill. An example of this would be the lesson with Mr. Turczyn. It was a math lesson. When you entered the classroom, your first impression might be that it was chaos. Meanwhile, they were just discussing. They argued, exchanged opinions. Mr. Turczyn presented some thesis which they refuted or confirmed. Yes, they were very creative children, all of them and very thoughtful. During one of the classes he tried to convince me that parallel worlds exist and he could explain to me how, in his opinion, they function in our human world. As for who he inherited his traits from, i.e. his mathematical inclinations, then without boasting, I think it was me. I was the class teacher and math teacher of the class that Wojtek attended. When he was in second grade, during the holidays, he came to me and said: 'I'm moving to Wrocław until the 14th.' Wojtek went, arranged it, and they accepted him. He also completed the third grade in Wrocław at the age of 14. There, he couldn't get along with his teacher, who was a Polish teacher. I think it's good that he came back to us from Wrocław, knowing that the Polish language does not belong to their world. My role will be to not interfere and allow students to pursue their passions. I prepared him for various professions, we worked together on math lessons, basically it doesn't matter. He had an individual teaching schedule, so he just absorbed everything, he was perfectly prepared and preparing these lessons often took longer than preparing all my other lessons, given the experience I had as a teacher. His great dream was to participate in the international mathematics olympiad, which took place in Vietnam in 2007. And to go there you had to take one of the first six places in the national Olympics. Well, one day before the final of these Olympiads, I had the opportunity to talk to an outstanding math teacher from one of the best top high schools in Poland, and he said that the passports to Hanoi had already been distributed, so there was no point in counting on the fact that my Wojtek would also fit in there. Well, it turned out that he took second place in that Polish general Olympiad. He won a silver medal at the Olympics in Vietnam and his path to all technical studies was open and he decided that at the University of Warsaw there was a program where they only accepted Olympians and he applied there and of course he got in without any problems. Later he decided that he would be in Paris at Ekol Poltechnik and defended his master's thesis there, because he had done his internship in the United States, at Nvidia, at Google. He came to the conclusion that he would pursue his doctorate in the United States at the University of New York, and where a doctorate normally takes an average of 5 years to complete, he finished, completed, and defended it in two and a half years. He usually came home for the holidays. We met many times for such a casual chat. I remember this dice he had, which he rolled and convinced me it made sense. Honestly, I didn't have to look for the meaning of it. I was in the middle of working on artificial intelligence, when it came to controlling cars, then on improving the kitchen. And one Sunday we arranged to meet in the city, we went to the Borek key to sit somewhere, have a coffee, he showed me the first foundations of artificial intelligence on his computer. He showed me how many servers to connect to, how it works, and how to ask questions. Somewhere in these various rankings, Elon Musk started to appear, and finally, he wanted to create a team of probably 10 nine. He himself indicated who he invited to join the team. Among others was Wojtek. And they started to act. He moved from New York to San Francisco. He still works there as one of the founders of Open Wojtek. He is independent and has his own opinion. It's not like someone is imposing their opinion on him. He has his own way of thinking, but he is very emotional from the beginning. A few years ago I received a kind of card from Wojtek, that is, a copied photo of us together and a text. Dear mommy, thank you for giving birth to me. And so on, for which I thank you. I will read a note that I received from Wojtek as an adult. I remember the moment when I got up from my chair in the classroom, walked up to the teacher and poked her belly with my finger. Thank you for giving me extra math problems back then. Wojtuś was and is a sensitive man. Let me read it. I hope he will forgive me. Hello. This is a poem from Wojtas, a great fool who often interrupted the lady and did not want to listen to her. But now, on the day of saying goodbye to this poem, I apologize to you deeply and wish you all the best. When you read this story, please remember our supposed peeing, how we climbed out the window to your vigil, or when I changed the home page to the explosive page. This is the end of my poem, because I'm going to eat something. Pierogi. Wojtek, I wish you only beautiful and good days. Be happy in the kingdom you create. I am very glad that this is the world that has been introduced here. However, it's also a great thing that you don't need the most top schools in Poland, you can also develop here in a place like Kruszbork. Now, if I were to appeal to young people, I would like to say that they live in a beautiful country, a country that is developing. Stay in Poland, work in our country, of course with the help of experienced people who play a very important role in the world. I am proud to have had such a student. He has surpassed his master and I am very happy about that. I wish him all the best, much success, and I hope that his work will contribute to us living in a better world. Regards, Wojtek. Wojtek, I wish you that your family will be happy with you and that you will be happy with your family, that your son will be your joy and your wife too, that you will enjoy spending time with him and be happy with him, and that they will give us even more offspring. Our sons went out into the world so that they wouldn't get lost in this world, so that they would be together on the occasion of various celebrations, family celebrations, holidays, so that they wouldn't forget about each other, so that they would be together.
W
Wojciech Zaremba14:21
Wow. I am truly grateful for the kindness of many people I have met here at TWKU, and there are probably many talented children who were not as lucky as I was to meet helpful people in their lives, who did not criticize them and gave them some level of freedom. And this happened to me many times, it happened to me with my own parents, with teachers, and along the way I also met many friends who were ambitious and allowed me to become who I am. Thank you.
M
Maciej Kawecki15:16
Wojciech, in this film there was a sentence that from an early age you had a question mark written on your face. It remains so to this day.
W
Wojciech Zaremba15:28
Yes. Well, it seems to me, it seems to me that fundamentally, even when I look at little children aged one, two, three, the child walks around everywhere and asks why this, why that, so it is even very natural for children to be curious about the world. And now the question is how not to kill it, how to make it constantly exist inside this person and not be afraid of it. It's like this, even in my life I've seen people who say: 'Oh, I've finished school, I won't have to study anymore.' This is kind of pointless. For me, the connotations are that science is terrible, but it's terribly interesting, and really, when you look at different types of elements of the world, if you realize, for example, you know, how passionately, I don't know, I read history and, for example, history describes what things happened and gives such a perspective, for example, what these things looked like in the Roman Empire, or what the beginning of humanity looked like, or even, as I read from physics, such an understanding of how the universe works. Well, quite interesting. Well, it's just that we function in a world that works differently than we would think. Yes? If you look very closely at what this universe is made of, it actually works differently than it might seem at first glance. So I can say, there is a lot, the world is more interesting than anyone would say, in any fairy tale anyone would invent it.
M
Maciej Kawecki17:17
What is the most important question mark for you today, Wojtek? If you had to choose one.
W
Wojciech Zaremba17:25
What is this topic that I am currently dealing with, artificial intelligence? And now artificial intelligence. A lot of people don't realize, and maybe we can talk a little bit about this, the magnitude of what we're talking about, and it's also such a transformative technology for the world that it's very, very important to focus and put enough effort into it so that it actually has a positive impact on the world. And there are actually many levels to making this happen.
M
Maciej Kawecki18:11
What is alignment faking?
W
Wojciech Zaremba18:16
So first of all, let me try, let me try to explain, I don't know how to translate it in Polish, what alignment is, and it's like this: there's a whole class of problems to solve. And this is also the case with many earlier technologies. So obviously when we built it there were a lot of problems too. For example, when we discovered electricity, electricity had this strange property that it was invisible, you could touch the cable and it would kill you on the spot. And so I can say that this analogy between artificial intelligence and electricity is not entirely correct, because artificial intelligence is at the same time both electricity, fire, a wheel, an atomic bomb, and a rocket. All things at once. R alignment is one, this is the goal of alignment. I don't know how to translate this word in Polish, but I would say such alignment, the purpose of alignment is to make the artificial intelligence system behave as its creators wanted it to behave. Now someone might wonder why this is even a problem, since we program these systems, why is it that some additional work is required for this? And maybe this, maybe this intuition about the fact that we are programming this is a little bit wrong. Artificial intelligence is a bit like growing a flower or raising children. Well, we are able to decide how much water we will pour in, how much fertilizer, how much light we will provide there, and we observe how this artificial intelligence develops. And in fact, these scientific studies are part of this, to often figure out how to adjust these parameters so that the plant grows very strongly. Now alignment.
M
Maciej Kawecki20:31
Once again, how is this before alignment faking, i.e. compliance compliance, compliance, compliance, compliance, because the point is why why am I asking about this too. Because today I want to show a painting by Wojtek Zaręba. Not from the past, but from the present. Now we have, let me clarify this question for our audience. Today, one of the greatest challenges of such fundamental developments in artificial intelligence is that complex models can carry out their own missions, which are in some sense detached from humans. Can I be more clear and explain what I mean?
W
Wojciech Zaremba21:21
In the case of this alignment, would we like to be in a situation where the system solves tasks in the way we would like it to solve, so that it solves these tasks, right? Uh, well, now you might ask, uh, what could possibly mean that this system doesn't solve tasks properly? And I can explain such simple situations within a human being when this could happen. Let's assume we have a man who is trying to maximize profits on the stock market. He is in charge of maximizing profits on the stock market. But at the same time, if he is to maximize these profits, we would like him to obey the law. Okay, there is Polish or American law, he should obey that law. And now from the perspective of this guy who plays on the stock exchange, it may be that he breaks this law a little bit and then he would be able to earn a little more. So there is this incentive, I don't know if this incentive is a Polish word, there is this incentive, and maybe if I really, really, really want to optimize it to earn the maximum amount of money, I might be tempted to bend this bravo. Well, now in the case of artificial intelligence, if we create situations in which there is a conflict between the goal and some constraints, we can end up in a situation where the model starts to cheat. And it is true that we are able to create these types of scenarios because this model very strongly wants to achieve this goal. Example such as people demonstrated. Whereas we are currently in the following place. We are at a point where models cheat and it happens sometimes. The main way models cheat these days is they pretend they did the job when they didn't. This is cheating by being lazy. This is a way of cheating. The thing is that we can build scenarios in which, for example, models are able to lie. This is a fairly rare situation, but it is very important to understand it at a fundamental level and be able to eliminate it. This is one of the areas that I'm currently involved in, because it also looks like this, that obviously, within artificial intelligence, there are actually many methods to make these models definitely, definitely, definitely smarter and at some point we will reach, we and other institutions will reach the level of human intelligence and also a level of intelligence smarter than humans will be reached. And now it's obvious that a lot of people, a lot of people, are starting to worry about this. This is justified. There are reasons to be concerned about this, but it is also the case that many people cannot even imagine how this could be a positive thing for humanity. I have been thinking about such analogies to be able to describe how the future might be different from the present. I have a few analogies. I'll try to say one here. We'll see if it appeals to people or not. But it's like this, if we look at most of the time that man has existed, man came into being 300,000, 200, 300,000 years ago. In fact, what seems normal to us today, such as getting up in the morning, drinking coffee or eating scrambled eggs, is actually such a narrow piece of time. And for most of history, people didn't live that way. Life really was very different. It is like this that most of the time we ran around the forest in groups and hunted animals. This is 95% of human existence. Even if you look back 250 years ago, most people, even most people in Poland, were farmers. They were farmers. 90% of the people were farmers and peasants. And it wasn't even known whether there would be enough food for the next winter. It was quite normal for children to die. It used to be that people often did not give names to their children because there was a high chance that the child would die. So it's like, at this point, we really think that what's happening is normal, but in fact, if we look at human history, humanity is skyrocketing. There is a huge change in how people used to live. And so I was wondering if we look at this period, we were also running in the forest in groups of 10 or 12 people. There, a person could have several items with him, because that was all he carried with him. He ran from place to place, hunting animals. He had quite a varied diet. Once we settled in, this diet shrank significantly. We started eating only grains and actually the moment we settled in, every single element of our lives changed. Our religions have changed, our culture has changed, our diet has changed, human genetics have changed, diseases have changed. We started living in groups close to each other and it was so bad that we started exchanging diseases. And now if we look at this change, actually every single aspect of life has been changed. And many times this can be said to be for the better, but there are also elements that are for the worse, as it turns out, for example, in the case of nutrition, when we were running in the forests, we had a much more diverse food. It was mostly uh proteins and fats. Now when we moved into agriculture it was uh, carbohydrates. And obviously in the past it took even thousands of years for such a transformation to happen, but there is also some sense that everything is accelerating. Speeds up. So 300,000 years ago our species arose. 12,000 years ago we settled down as farmers. Now the Renaissance is 600 years ago. This industrial revolution 250 years ago. Computers were built 60 years ago. It started working 10 years ago, so everything is compressing, compressing, compressing, and it's like this: we are, this is a technological development that, to some extent, maybe stems from human nature, that we are so curious, that we want to understand how nature works, we want to make life better for us. To some extent, this is technological progress that is unstoppable and will change every single part of our lives. I think so. Yes. Well, everyone will change this and this will also change many assumptions about what we consider normal and what we consider abnormal. It is so that in the future, I read in a book that it was believed that it was from us, that war, famine or disease were something inevitable for man, that man was simply one of these three things that he simply had to constantly be throughout his life. And it turns out that, for example, in the field of food, we were able to invent such techniques that currently 2% of people in the world are engaged in agriculture, are able to produce this food so that if we wanted, everyone could have food. We somehow decide within the system we have that maybe not everyone has food, but we are able to produce food for everyone. It's not a lack of technology for food production. Um, we've also made quite a lot of progress in terms of diseases and it's really a whole range of these inventions that you take them for granted and you assume that they're normal and that they've always been there. Well, it's like this, for example, we have chlorinated water in our taps, all over the world there are these waterworks built, there are pipes running through cities. There is water that is chlorinated and therefore does not have bacteria. There is a habit introduced everywhere in the world about washing your hands. Somehow we were, we were able to convince everyone in the world that it made sense. You should wash your hands. There is soap in every toilet. We also built toilets, we built antibiotics. We have also invented vaccines.
M
Maciej Kawecki31:15
Has artificial intelligence quietly become an intelligence equal to human intelligence?
W
Wojciech Zaremba31:20
I don't know what equivalent means. It is capable of performing individual tasks at least as well as a human. Whether it's because it can't yet perform complex tasks, I don't know, it won't run the company for another 10 years, we know, but if we look at individual executions of individual tasks, has it already surpassed human intelligence in that single aspect, or is it something that people have thought for some time, that was the main way of thinking within this artificial intelligence? I was wondering which jobs would be automated first. Okay, there was this way of thinking and it seems to me that we are closer to the fact that this artificial intelligence within each job, maybe within some more, some less, is in such a way that it solves some tasks and I kind of imagine that it is, there is this terrain of mountains, there are hills and and and hills and it is filling with water and slowly filling with water, that there are more and more of these functions. Artificial intelligence is able to help, it is able to solve. And now we're in a place where I'd like to talk a little bit about the possibilities before I talk about some suggestions, just to give people some perspective on how much technology we're talking about.
M
Maciej Kawecki33:22
Why am I asking you this? I'm asking you in this specific context, so that I may play a bad cop now, but to very, very vividly show how fundamental it is today to stay up to date with AI developments. I met several times with Jeffrey Hinton, who you know very well, this year's Nobel Prize winner, a father, one of the founding fathers of artificial intelligence. He actually says that man will become the second, not the first, intelligence on Earth. Whether we agree with this or not is a secondary matter for me, but if we were to assume that artificial intelligence is equal to human intelligence in the implementation of first one task and then a complex number of tasks, then in the history of humanity that you mention, we have only had one case where two intelligences were at the top of the food chain and one ceased to exist. And that was the situation where we had Homo sapiens and Neanderthals. The question is, can we use this metaphor to describe today's reality and the situation of people who say: 'I'm saying no to artificial intelligence, no, I won't learn, I won't educate myself, I won't use AI tools. Could they end up like those Neanderthals? Would I rather? Or are we entering a period of time where artificial intelligence can help a given person 5%, 10%, 20%, but this will increase. It will be that a given person will be able to do something that 10 people can do. We will have companies where a small number of people can create a multi-billion company.
W
Wojciech Zaremba35:20
And it's like this, it results from many things. It results from the fact that if, for example, I wanted to create a company today, I need, I don't know, to take, to find a financial investment to hire people, and people are quite expensive, and so a person can have someone to help with the law, to help with the co-founder of the company, so really there will be shifting tectonic plates regarding what future company structures will look like. There's a huge number of them, I mean, people have many needs, which is great. People would like to be healthy, for example, they would like their children to be educated, they would like to travel around the world, see different things. They would like to fly to another planet. There are many needs that people would like to fulfill. So there's room to build thousands, if not millions, of different kinds of companies. And these are, and in fact, artificial intelligence can help an incredible number of people produce all kinds of great goods. And to some extent, it remains to be seen whether this will happen, as I said, it reduces the barrier to entry, to creating businesses, and I can say, it's quite exciting. It's a bit like yesterday's Elevenlabs, for example. There's a company from Poland called Eleven Labs. It's currently in existence for 3 years. Within two or three years, this company's valuation is $3 billion. Yes. A huge, huge amount. It's surprising how quickly these companies are able to generate such value. And now, um, if I were to ask, I don't know, if someone were to ask me what the current skills are within artificial intelligence, whether this artificial intelligence will solve everything, I would think so. If I think about it from a company perspective, for example, even if the company has super-smart employees, it's still valuable for the company's CEO to be well-versed, to have a deep understanding of how the world works, to be able to manage them well. So, more or less, that's the direction we're heading in. So, to some extent, first of all, it's very valuable for people to learn, but to learn with understanding. To learn so that they understand fundamental things as deeply as possible, not 100 things very superficially. It's not a matter of memorization. And there's another skill that I think is very, very valuable. It's the ability to truly utilize AI to its fullest. Right? Well, it's going to be a large group of people, I use it occasionally. And I can say that this additional skill is generally figuring out what maximum things can be done within all the technologies that have been built, or even building more complex technologies than what exists now.
M
Maciej Kawecki39:04
What do you think the next few years of artificial intelligence development will look like? What will we get from the world? I'm not asking about Open AI, I don't want to ask about what awaits us from the Open AI company. These are your inner secrets. I'm interested in, I ask a little about products, but in isolation from Open AI. Will we be looking at multi-task AI, meaning we'll suddenly have agents that will manage our computers, help us book flights and hotels, or maybe the world will move towards scientific AI, meaning we'll be creating copies of Alpha Fold and Deep Mind, also awarded a Nobel Prize this year. What can we expect from the development of artificial intelligence in the coming years? I'll answer this question, but I'd also like to share a story from a friend. A friend had a very complicated relationship with his wife. The birth. It was a 17-hour labor, quite risky. There were even two shifts of doctors during the labor. They didn't quite agree on what to do. My wife was given all sorts of medications, and during the labor, he was constantly with his phone, entering everything into the best Open AI model, and the model made various suggestions. This colleague made these suggestions to the doctor. First, the doctor laughed at him, and after a while, you found it on Google. 'That's interesting.' And it happened that the doctor actually tried several of these suggestions. One of them was to do an ultrasound, and it turned out the umbilical cord was around the baby's neck. Yes. Now, it's unclear; perhaps the doctor himself would have come to that conclusion, but it's quite incredible, if this technology can help in such situations at this point. And that's not the only story. I also heard the story of a woman who went to the doctors and wasn't feeling well. The doctor told her: 'Maybe you should change your diet.' She went home, ordered some lab tests, and also talked to the model. She sent them to the model. The model asked her if she had the lab tests. She sent the tests, and the model wrote in capital letters that she should go to the hospital immediately for emergency treatment. She was a bit surprised. Maybe this was some kind of scam. She looked at the lab test. At the very bottom, it said it hadn't been checked by a doctor yet, and it was the weekend, so it wouldn't even be checked for another few days. She went, took a taxi, went to the hospital, and they tested her on the spot and told her: 'It's true that her platelet count is dropping to zero.' If she hadn't come, she would have been dead in a few hours. And the doctors were shocked, because normally people who are at their best don't go to the doctor. It's getting to the point where he actually loses consciousness and needs to go to the doctor. And you asked her how, how did you know to come? And she says: 'I actually talked to this model, and this model suggested it to me, and then the doctor said, 'Well, in this situation, I support this artificial intelligence.' So I think what do these stories say about the future? Do these stories say there will be a transformation of medical services? It's true that we can be definitely healthier, and I'm very
W
Wojciech Zaremba43:14
I'm excited about it. This will obviously develop for some time. There are also a lot of regulations, and there's also the question of what if the model says something false? What should we do about it? These are significant problems. However, it's quite incredible to be in a situation where a person has a phone, takes a tiny device out of their pocket, and this tiny device has the condensed knowledge of many doctors, scientific papers, research, is able to answer questions like that, and quite often they answer these questions correctly. And now, for example, we're even trying in Africa, in hospitals, and we applied ChatGPT, we changed this infrastructure in hospitals and it is actually able to significantly reduce the number of errors. I think that over the next few years, it will be quite obvious that healthcare should function differently. And it will function differently, and it really depends on the countries' readiness to implement such solutions. There are countries that are more ready, for example, Singapore or even African countries where there is, for example, one doctor for the entire country, so no one there is afraid of losing their job. It's worth mentioning that Poles, working with the well-known Yann LeCun, created the most effective algorithm for diagnosing breast cancer at New York University. Indeed, the statistics of medical imaging data analysis there, relative to this organoleptic analysis, are astonishing.
H
Host45:03
Wojtek, and you'll ask me about the fields. I told you, I only told you about medicine.
W
Wojciech Zaremba45:09
Medicine. It's also true that I think there will be significant scientific progress in artificial intelligence. This is partly due to this, if you think about any field. Maybe it could be cancer, it could be quantum physics, it could be inventing new special materials. It's like this: within each of these fields, you might have 10,000 specialists in the world, right? 10,000 people who were able to dedicate their lives and understand a given field very, very deeply to such a degree that they are able to invent new things within it. And now, if we are in a situation where we can support them with models that have very similar skills or even better skills, then suddenly there will be not 10,000, but a million specialists in those fields. So various scientific fields will advance. And that is how it is. For example, I like to look at the world through the eyes of a child. Even when I think about the current materials we have, for example, an ordinary material like glass. Glass is melted sand, and it's a material with incredible properties. It's transparent, it can hold water inside. We've mastered the technique of building glass. And if I think about it, about those guys 12,000 years ago who were running through the jungle, if they saw glass, well, that's a cosmic material, right? It's unbelievable that something like that could exist, right? Well, there's nothing like that on a tree, nothing like that grows, you can't find anything like that underground. It's a completely human-made material, and we've created a ton of these materials that have incredible properties, and I think we've invented many more such materials. And even if you look at individual materials, so far they could have had such a significant impact, that we even called the entire period of time the Iron Age, the Bronze Age, that's one invented material, right? And I think we'll invent many more materials with incredible properties, and I think we'll also make tremendous progress in the treatment of all types of diseases. And that will happen in the coming years. I think so. That's right. So if I think about medicine, there's also the problem that testing on humans takes a relatively long time, meaning inventing new drugs takes about 10, 15 years, which is a certain limitation. Currently, we have 4,000 drugs invented, so I can say that you can have geniuses in the computer, you can have them in this data center, you can have Einsteins inside. However, the world also operates at a certain speed. In some areas, faster, in others, slower. So, for example, in software, you can write software very quickly and upload it to the internet, and a million people see it at once. In the case of drugs, it's a bit different. Even if someone invents this drug, there are animal studies. The first phase of human trials, the second phase of human trials—the first is toxicology, the second is efficacy, the third is large-scale human trials, and that simply takes some time.
H
Host48:53
I have two final questions for you before we give the floor to our audience. The first is whether, in your opinion, artificial intelligence should be a mandatory subject in school?
W
Wojciech Zaremba49:06
I can say, that would be truly impressive, and it's the right decision to move towards teaching children how to use artificial intelligence. Obviously, you need to think carefully about how to do this, because artificial intelligence is so fundamental. We also want to teach children to understand it deeply. Yes. And it is possible at this point, and that's the problem in schools and universities: you can ask artificial intelligence a question, it gives you an answer, and that's it. And the child doesn't understand. So now the question is how to build this program, how to do it in such a way that it truly ensures that the child fundamentally understands these fundamentals of science, because they turn out to be useful throughout life, and at the same time, in the best possible way, so that they can use this artificial intelligence to build new things, invent new things. And in my opinion, that's great, right? If Kluczbork or Poland were able to do something like that, in my opinion, it would be a very good investment. However, you have to seriously consider how to do it so that it doesn't actually rob children of their critical thinking skills. An interesting thing is that, for example, in China, when it's the annual high school leaving exam, artificial intelligence is turned off or degraded across the entire country so that children can't use it. So it's interesting that even at the national level, decisions can be made that actually maximize the positive impact, not the negative. And even within China, they've also made smart decisions within these social networks. TikTok in China is definitely more educational. In the case of the Western world, technology has been built that is incredibly addictive for children and has a somewhat limited use, but it often has a negative impact on children. I hope that artificial intelligence won't have such an impact, and at the same time, I can say that artificial intelligence is a general-purpose technology. It can be turned into a technology for healing people, for education, for solving the world's biggest problems, and also from this same technology, it can be turned into a technology that would be the most addictive. Yes. And many other things can also be used to develop weapons and so on. And I would probably expect it to be used in every single one of these domains. It's a bit like a computer, right? A computer is used for good and for bad purposes. And we should expect the same with artificial intelligence. And now I would say that a valuable approach, at least that's how we've always approached it within OpenAI, is firstly focusing on what the potential possibilities are and secondly asking ourselves how we can minimize the negative consequences. And this way of thinking allows a person to move forward, to start building. The second perspective, which mainly focuses on the negative aspect, causes paralysis and causes no progress. I remember yesterday we also had such a conversation about what he was talking about, I don't know to what extent I can have any influence on it. The fact is that Poland has an incredible culture. It's beautiful. A country, a huge number of smart people. And now, if I'm wondering what the difference is between Silicon Valley and how people in Poland think, it mainly comes down to the fact that in Silicon Valley, people are more optimistic about the future.
H
Host53:54
That's true. What can we build? What great things do we have? Yes, we have a Polish white-tailed eagle. We have many Polish things. Well, I wanted to ask you at the end about Poland. Because we did create a language model, a rather large white-tailed eagle, which, despite everything, is hard to compete with your ChatGPT. What about Poland? Is it? Yes, I was thinking, have we already lost this race in some sense?
W
Wojciech Zaremba54:34
Even as I look back on my life in terms of competitions or races, I remember my mom remembering that I once entered a competition and somehow failed and came back home. I won't worry too much about it. I just started doing more math problems.
H
Host54:51
Are you saying that we haven't been successful in this competition yet?
W
Wojciech Zaremba54:52
No, I'd like to say that... The real thing to focus on is what to build, not whether I'm the best or the worst, Wojtek. There are tons of things, because there are actually tons of possible things to build. You can build all sorts of companies in every single field, and you can apply artificial intelligence. I've already mentioned it in medicine, education, and really in every single field. It can be applied to fashion. You can think about how to make transportation look different. You can apply it in every single field. Once the question is there, it may also be that in certain fields, AI is not sufficient at the moment, or other aspects are perhaps the most important, but that will also change. This intelligence will be more valuable. So I can't answer for every person what the most valuable thing is, what a person can do if they believe in themselves. Yes, it's not even a question that if I gave the answer, it would have the right result. Yes. Fundamentally, it has to come from within. A person must be curious themselves and must be. So, it's a matter of asking. What interests me at all? What's interesting, what would I like to see in the world? Yes. And I think everyone will find the answer to this question. There are plenty of answers to this question. And if not, you can ask ChatGPT.
H
Host56:37
Thank you for this meeting. I've been feeling the pressure for 10 minutes to give the audience a chance. This is absolutely the time. Ladies and gentlemen, a very inspiring meeting, but I promised you there would be questions from the audience. It's amazing that you came here to meet locally, and not just at large conferences. Thank you. On behalf of everyone, I wanted to ask in detail, but this is a large conference. I see it more as a meeting. I wanted to ask in detail about the aspect you mentioned about clinical trials, that artificial intelligence will somehow accelerate it. Do you see how exactly this can be accelerated, considering that these human trials are now such a total standard? Are there several elements?
W
Wojciech Zaremba57:33
It's like, as part of research, or as part of drug development, the first element is to propose what molecule it is to be tested, and this is the part where AI can definitely help. The second element is animal testing. However, within the animal model, it's also a matter of understanding the analogy between animals and humans. Within this framework, animal models are close to human models. Now, within this framework, within human trials, it may also be a matter of how regulations exist. I've heard that some companies, instead of conducting research in the United States, are starting to conduct research in China. Obviously, there's also the question of how to achieve regulations that make sense, are safe for these people, and can compress this time. In my opinion, the question of how to change these regulations in general to be able to speed up this time is very sensible. I was also surprised, for example, that once you go through these three stages in the states of human testing, apparently it typically takes six months to prepare and submit the application for the drug itself. So I wouldn't even have realized that these six months are something that can definitely be avoided. Compress the drug's application time to a much shorter period. And there's actually another approach. Companies are building new drugs at this point, because new drugs can be patented, and once a drug is patented, you can definitely make money from it. However, if the goal isn't just to make money, but to help people, it turns out that among the existing drugs, 4,000 drugs, we've repeatedly seen that drugs can help with drugs whose patent has already expired—meaning drugs on which the company can't really make money. These drugs can help with diseases that haven't been used before. So if this is more from the perspective of, for example, the Ministry of Health. If the Ministry of Health or some donors would like to solve more diseases as a goal, then you can actually build some kind of investment method or some kind of financial method and financial instrument that allows people to look for drugs like the ones we currently have, which have been tested and are safe for people. Find their applications for diseases where they haven't been used. If that were to happen, it would be possible to shorten the time from 10, 15 years to one, two years. So there are many steps, many possibilities for accelerating this cycle.
A
Audience Member59:00
Please.
Good morning, Damian Czajka, sound designer, composer. It's very nice to meet you, and I've enjoyed listening to all these conferences here. Also, the presenter. It's a pleasure to watch the channel. So thank you for your professional handling of this conversation. I also have a question about building. Building here in Poland. I have an idea for a startup to create libraries for feeding AI algorithms, for feeding quite specialized algorithms that would analyze sound or create sound, for medical applications, for security applications, even, unfortunately, military issues. And above all, for entertainment, the gaming industry, the video industry. What specifically would I like to create? Very, very extensive libraries that take into account all possible articulations of sound intensity. Recording at very high frequencies, and currently the highest possible, i.e., 384 kHz. Recordings on special measurement microphones up to 100 kHz, much higher than the human ear can hear at 20 kHz, because by resampling this sound to our hearing level, we obtain extraordinary cosmic sounds, which today we can treat as people from the past seeing glass. My question is, where should we approach such a startup? Should the model be open source, meaning shared with other structures that create models that would like to gain access to such sound databases so they can have them open source, or should we create a business model or a mix of these models? Where exactly should we look for investors? Where would you say is the best place to take the project?
W
Wojciech Zaremba1:03:07
So, within any company, there are two typical stages. There's the first stage: the idea, trying different things, trying to find what works, what people are interested in. There's the second stage, when we actually know there's huge interest in what we've done, and then it's just a matter of doing more and more, because people are willing to buy the product. And now finding this is called product market fit, or how would you describe it? Did he translate? Weren't you listening? I wasn't listening. I was currently looking at your... I was currently looking at your little one. Well. Okay. There's an explanation. So, at this point, you're asking me what needs to be done to assess the market, to arrive at this product, what needs to be done to know that what you're doing makes complete sense. And I can say, well, it's hard for me to say, each of these things you proposed sounds sensible. Well, I don't know exactly how your market works, and it's probably like this: you have to talk to many different people who might be potential customers and even be able to say, 'Listen, I have something to sell, you're close to making it.' Maybe you don't even have what you're saying you want to sell, and you just want to see what the reaction is, whether these people are willing to pay for it. So if it's like this, in the first phase of the company's existence, if you don't achieve this product market fit, this market validation, then the company might die. So, it doesn't really matter whether the idea sounds great or not. You know, it's really the market that verifies whether what you're doing makes sense. And the test is very non-trivial. At the moment, if, and as a standard practice, at this first stage, you try not to hire too many people. At this second stage, when you know it's working, then maybe you should hire more people to sell. Some entrepreneurs make the mistake of suddenly hiring more people, it's proof that the company is working, so they hire more people when the product isn't even verified, and that in itself causes the company's costs to go up, and the company is failing. So at this point, you have to see if what you're doing actually has a demand. Maybe there is, maybe not. And I think we can afford to reflect that the world isn't exactly moving towards open source solutions. I can say, it depends a bit on whether there are many types of companies, masters of different kinds, you know, for example, but whether it's not like, I won't mention them by name, although most of us know which one I'm talking about, the largest one, but it's not like the largest companies are creating open source solutions primarily to weaken the competition, right? It's also like, for example, let's say Google took, invented a method, a way to perform large calculations using the CPU. It's called Map Reduce. And when they did, they created internal software that hundreds or even thousands of programmers use, and that's not their main product. It's better for them when this product becomes public and people outside learn how to use it. It's even easier to hire people to use it, than if someone else did. He would create a product that is similar to Google, but not exactly like Google, and then it's not like that when they release it, they get it. At the same time, people like them more. People are grateful because you're releasing technology for free that cost a million or hundreds of millions of dollars to produce. So many people appreciate it, and whether or not, as I could say, there are different situations. Maybe sometimes it's about weakening opponents, but it really depends on the situation. It's complicated.
A
Audience Member1:08:07
We're asking you this. Um, Wojtek, I represent the National Revenue Administration and I wanted to tell you that we're very proud of you, because Wojtek, even the National Revenue Administration likes you. Because you're a scientist, you're a visionary, but above all, you're a wonderful person guided by values and for whom the well-being of humanity is important. And my question concerns the possibility of implementing solutions of such an individual model in our national tax administration, and it's not just about communication with our clients, but also about whether we all want to. I think yes, I think yes, because it's not just about communication with our clients, but precisely so that we can... We should create a whole so that we are all happy, so that our customers and our taxes remain the answer. Please or thank you,
W
Wojciech Zaremba1:09:10
I think it would be incredibly good if the country could implement various types of AI solutions within the framework of how government institutions operate.
A
Audience Member1:09:30
Is this a government institution or not?
W
Wojciech Zaremba1:09:31
Yes, yes, public. And it's true that often when we look at these government institutions, they are slow, people don't like to use them, and I can say there is some opportunity to move forward at this point, and there are a lot of these solutions. You need to create a team that deals with this, a team that could even figure out where the greatest value can be added. Yes. And it's a bit like building a product, building a product, looking to see if people use it, whether it actually accelerates it, reduces the number of problems, and in my opinion, that would be very, very wise, and I can also say that it is even motivating. It's true that within the people working in such an institution, if suddenly there is an approach like, listen, we want to use the most advanced technology, whatever it is, we want it. There's a chance right now, there's a chance we can build better institutions than the institutions that have existed so far. And we need to build a team of ambitious people. I think that would be great.
H
Host1:10:58
I don't know how much time I have left for questions. Last, last. I think the last question. You can sit in the front row here, because so many times. Ladies and gentlemen, there could be thousands of questions. This conversation is very inspiring. I have to keep an eye on the time. Mr. Wojtek's time is very valuable. His family is also waiting, so here and there, and we move on to the next point.
A
Audience Member1:11:22
Wojtek, first of all, congratulations on your passion, talent, and your entire story. You mentioned, Maciek mentioned Poland, Bielik. The room responded with smiles. And I'd like to ask you about your opinion on the role of local language models and the work of those people sitting here in the fourth row, because they are part of the team that is creating Bielik, the Polish Bielik, completely free of charge. Raise your hands, where you are sitting. Yes. And I'd like to ask you if you could share not only your opinion on these roles, but also a few words of encouragement and positive energy for these people.
H
Host1:12:10
That was the plan, of course. Is it really possible to look at it, and I don't know if Wojtek even knows that Bielik is a social act. Great. People did it without any remuneration.
W
Wojciech Zaremba1:12:26
It's possible to look at it from several perspectives. One or the other. Working on a project like this, if you think about it, even regardless of whether it's better than the GPT greyhound or not, leads to the team beginning to understand how this technology works, and that opens up opportunities to build various types of technologies. Perhaps it makes sense, perhaps something a little closer to what has what, something that will have product market fit and a huge impact. However, it's a bit like that. I heard an analogy. Imagine there's a coliseum, and now there's an audience in the coliseum, and there are gladiators in the middle. And sometimes people in the audience laugh. I say, 'And this gladiator here got hit in the face.' They're laughing. I can tell you, those people in the audience don't interest me much. I'm more interested in those other gladiators. And now you are gladiators. You're doing, building, and now there's one question: does it make sense, or how to do it in the end? That's a question for you. What should be done to achieve maximum success, to make it the most valuable? I don't know the exact answer to that question. At this point, you're approaching the fact that you're building, thinking, trying to understand this technology, and in my opinion, it's very important, and I would ask for a round of applause for the team. If people who don't do anything criticize you, there's no point in listening to them. How they do, how they build, is worth listening to, what they don't like, but if they sit on the sidelines and do nothing themselves, there's no point, there's no point in listening.
A
Audience Member1:14:41
Jarosław M, Technology and Science Park in Opole. Thank you to the organizers, but especially to the gentlemen, for a very interesting and pleasant conversation. However, thank you very much, Wojtek, for what you said at the very beginning, that you're fighting for this in Open, so that the program operates legally, that is, within certain limits. But what happens if every model has certain boundary conditions? Yes, what happens if this model begins to slip beyond the boundary conditions and moves towards the dark side? Yes, we say great, you're presenting these elements that serve humanity, yes, but can't these things be used against humans? Is this technology so broad that it's practically certain that it will also be used against tech, against humans?
W
Wojciech Zaremba1:15:23
It's the same as computers that are used, I don't know, in banks or on some websites for search engines; they're also used in drones, right? And within the framework of artificial intelligence, it will also be the case that, for example, biological weapons will be invented using this technology, or how, I don't know, developing nuclear projects, or generally how to invent new weapons that haven't been thought out so far, it's practically certain that it will go in this direction. And now, when I think about it this way. In society, we have criminals, yes? And it's almost inevitable. We'll have some people who have a negative attitude, who want to have a negative impact on society. And right now, it's not like that. Despite the existence of such people, we still have a relatively stable society, and we have a stable society because, for example, the police exist. Yes. And this also results from the fact that most people, 99% of people, are not criminals, they only try to do something that is nice to other people to some extent. People are mainly altruistic, they want to do something that is nice to other people. So I think we'll be in a very similar situation, with a mastery of various models, a mastery of various applications. Some of them will be positive, some of them will be negative. And now it's very valuable for these positive applications to be the first, larger, more significant. And that's what I'm appealing for.
H
Host1:17:28
Thank you very much, Mr. Wojtek, Mr. Maciej, for being so inspiring. I just wanted to add one more word at the very end: I've been watching Wojciech for two days, spending time with him, and first of all, I see, because we've known each other for a while now, I see his enormous evolution, and that he's paying attention even more than before. I'm focusing on this human factor, but I also wanted to share a reflection of my own: I organize a lot of different events and I haven't yet encountered a story where, in such a short time, such a huge number of people were interested in meeting anyone, especially considering that when organizing the meeting at the Jagiellonian University, we set the strictest possible criteria, meaning only winners and finalists, only and exclusively from STEM Olympiads, and only and exclusively from this year. And despite this, we managed to gather an audience of 600 people in a few minutes. So I told Wojciech that probably in a year and a half we would organize a meeting for talented Polish people at the National Stadium. So, Wojciech, I wanted to thank you very much for these two days. I think it has a very big impact on Poland. I don't think you can even imagine how much. Thank you.
Ladies and gentlemen, I'll add to the Kluczbork thread. How we started distributing free tickets in two hours at the Bajka cinema, how we increased it to 500 people in the OSIR hall, and the exit tickets were already sold out. Ladies and gentlemen, we now move on to the ceremonial part. We would like to honor Mr. Wojtek Zaremba with a 'Kluczbork' tower. I invite the mayor of Kluczbork, Jarosław Kielar, deputy mayor Roman Kamiński, and the chairman of the city council, Mr. Piotr Sitnik, to the stage.
N
Narrator1:19:39
[Applause]
M
Mayor1:19:42
Wojtek, ladies and gentlemen, dear family, dear residents of Kluczbork, but residents of all of Poland. You influence the life of Poland and the world. And here in Kluczbork, we have a symbolic tower. This tower, which we are giving to you for the second time. Two years ago, your parents received it, one might say for the JPA chat. Now I would like to give you this tower not only for your education, but for your heart. You showed how big a heart you have last year by donating a considerable amount of money for the development of education at the high school you attended, which is just across the street. And importantly, you also donated money to young people who will want to educate themselves, develop themselves, and perhaps create new works. For all of us. For this great promotion for Kluczbork, we would like to present a symbolic tower, and I hope it will find its place somewhere in San Francisco. Thank you and congratulations.
H
Host1:21:06
The mayor mentioned the high school from which Wojtek Zaremba graduated, and the Complex of Higher Secondary Schools would also like to present a special distinction to one of its most outstanding graduates. I invite representatives of the first high school in Kluczbork, headed by Principal Joanna Piłat.
P
Principal1:21:33
Good morning, ladies and gentlemen. On behalf of the entire school community of the Complex of Higher Secondary Schools in Kluczbork, I would like to present Wojciech Zaremba, our graduate (you already know he's here), with a special award. It's the Adamus Award. For several years, we have been presenting this award to students who have distinguished themselves not only with their grades, but above all, with their activity and contributions to the school and local community. Now is a somewhat historic moment, because for the first time, I will be presenting this award not to a student. It will be in the 'Friend' category. I would like to do this in the company of Mr. Piotr Ostrowski, who represents the Mickiewicz Educational Association. This association implements projects that we can implement thanks to our cooperation with Wojciech Zaremba. Mr. Piotr Pawlikowski is also with us; you've already seen him in the film today. He's a mathematics teacher and Wojtek Zaremba's tutor at our school. Today, he's also representing the jury of the start-up scholarship funded by Wojciech Zaremba for our most outstanding graduates who are starting their journey to higher education. I regret that Headmistress Krystyna Włażewska, who initiated the collaboration with Mr. Wojciech Zaremba, is not with us today, but I am convinced that she continues to patronize this joint venture. And to make this even clearer, so that you can better understand why Wojciech Zaremba is receiving today's award, I would like to read the text of the diploma that will be attached to the statuette. By decision of the jury of the Adamus Award, the title and Adamus statuette in the category of 'Friend of the Adamus Secondary School Complex' Mr. Wojciech Zaremba receives the Mickiewicz Award in Kluczbork. For being an inspiration to young people, promoting the school's good name, and providing genuine support in creating a space where talented students can spread their wings. Congratulations and a big round of applause.
N
Narrator1:23:54
[Music]
H
Host1:24:51
I did my best. I think this meeting was, above all, an inspiration, and for the young people, hope that you don't have to be from Warsaw, New York, or Paris to achieve success. All you need is a small town, a good family, good friends, and good teachers who will guide your genius in the direction Mr. Wojtek took. Thank you for your attention. See you soon.