A Pole quit his job at OpenAI and created a business worth $4 billion. “I want to open an office in Poland”
– In the near future, my company will open a representative office in Warsaw – says Jarosław “Jerry” Tworek. A Polish scientist left his job with Sam Altman to start his own company. Core Automation achieved a $4 billion valuation in just a few weeks.
Szymon Krawiec, “Wprost”: Why did you leave OpenAI?
Jarosław “Jerry” Tworek: I have worked there for the last seven years. At that time, OpenAI became quite a well-known company in the world. But it wasn’t like that at first.
At OpenAI, they had an idea – to focus research towards general artificial intelligence and create large language models. And they were right, because they became a promising direction, and OpenAI was the most innovative company in the world in this respect. It could create new models every year to eventually become the leader in the artificial intelligence market. And during this time it has grown to enormous proportions, becoming the largest economic force in the world.
But as the OpenAI market has grown, so has the competition. Especially with Anthropic. At this point, this rivalry is reaching extreme levels. So OpenAI began to focus on this competition, and its ability to innovate, make big changes, control and release new language models, unfortunately became somewhat limited. This is how it is: success, on the one hand, gives you fuel for growth, and on the other hand, it also narrows your field of potential action and possibilities.
If someone believes, as I do, that current artificial intelligence solutions are only the beginning, not the end, then it seems to me that the best solution is to leave a large company that is already a bit ossified and structured and try to do something new on your own. It is important to take into account that the field of artificial intelligence is still very young and there is still a lot to do and figure out.
In your farewell letter, you wrote directly that you wanted to “explore the types of research that are difficult to conduct in OpenAI.”
Yes it’s true.
So what research of yours did OpenAI not want to develop?
