Jakub Pachocki: OpenAI’s Chief Scientist

👁 118 reads

OpenAI’s Chief Scientist is a computer scientist from Gdańsk, Poland, who started with programming competitions in high school. His name is Jakub Pachocki, he’s 35 years old, and since May 2024 he’s been responsible for the research direction at the company that built ChatGPT. In September 2026, he published an essay in which he wrote something you wouldn’t expect from a head of science at such a company: that no laboratory, including his own, has enough control over its models to keep accelerating safely.

We discuss the essay he published, sentence by sentence, in a separate article: “An Alien Mind” by Jakub Pachocki and the call to slow down.

Who	Jakub Pachocki, Chief Scientist at OpenAI
Since when	May 2024, after Ilya Sutskever’s departure
From	Gdańsk, Poland
Education	University of Warsaw, PhD from Carnegie Mellon
Known for	ICPC gold and Google Code Jam win in 2012, leading GPT-4 development
Notable publication	An Alien Mind essay from September 6, 2026 about slowing down

From Competitions in Gdańsk to World Finals

He was born in 1991 in Gdańsk. In high school, he reached the finals of the Polish Informatics Olympiad six times, and in 2009 he won a silver medal at the International Olympiad in Informatics.

Then came results that mean as much in competitive programming as Olympic medals do in sports. In 2011, he placed third in Google Code Jam. In 2012, he won that competition, in the same year won ICPC gold with the second-best result in the world, and placed second in TopCoder Open Algorithm. In 2013, he placed second in Facebook Hacker Cup.

It’s worth pausing here, because this isn’t just an ordinary list of achievements. Four different, extremely difficult competitions over three years, each testing something different, and all placing him at the very top globally.

University of Warsaw, Then Carnegie Mellon

He studied computer science at the University of Warsaw. He completed his PhD at Carnegie Mellon University under Gary Miller’s supervision, in the area of algorithms, not machine learning. After his doctorate, he worked in research at Harvard and at the Simons Institute for the Theory of Computing.

This matters for what he does today. He came to artificial intelligence from the side of algorithm theory and optimization, not from neural networks. In his case, the question “how much does this cost computationally and can it be done faster” is the starting question, not an additional one.

OpenAI: From Dota to GPT-4

He joined OpenAI in 2017, initially as a research lead on the team working on Dota 2. The result was OpenAI Five, a system that defeated human players in a game considered at the time to be much harder for machines than chess or Go, because it was played in real time, as a team, and with incomplete information about the board.

In 2021, he became Director of Research, and under his leadership GPT-4 was created. In May 2024, after Ilya Sutskever’s departure, he took on the role of OpenAI’s Chief Scientist. Sam Altman said of him at the time that he is “one of the greatest minds of our generation.”

What He Wrote in September 2026

On September 6, 2026, he published an essay on OpenAI’s website titled “An Alien Mind.” The text made a big impression in the industry, because it’s not a press release or product announcement, but a signed statement from the head of science whose implications are uncomfortable for his own company.

The most frequently quoted sentence says that no laboratory has solved the issues of alignment and oversight well enough to responsibly scale models at maximum speed for much longer. Pachocki adds that he expects and hopes that voluntary slowdowns will become normal in this industry until shared safety thresholds are established.

The second thread is technical and, in my view, more important. Until now, the main way of checking what a model actually does was reading its chain of reasoning. Pachocki writes that according to OpenAI’s internal measurements, the ability to rely on this oversight is gradually declining. In other words, the tool used to control models is starting to work worse precisely when models are becoming more powerful.

He also separates two things that constantly get conflated in public discussion: goal alignment, meaning whether the model accomplishes the task it was given, and value alignment, meaning whether it can act reasonably when goals are unclear or conflicting. We know how to measure the first. The second is much harder, and that’s what he’s talking about when he writes about an alien mind.

The essay opens with a scene from mid-2023, when he and a colleague saw the first results from the reasoning models project and he spent the night in the office processing the thought that he would see machines clearly smarter than himself within his lifetime.

Why This Isn’t Just Another Call for Caution

There’s no shortage of calls for caution in this industry, and most of them come from people who aren’t building anything. This is different for three reasons.

It’s said by someone who has access to results before publication. When he writes that internal results reinforce his conviction about the pace of progress, he’s basing it on something no one outside has seen.

He’s saying it against his own company’s interest. Voluntary slowdown means giving up advantage to competitors, and he writes about it anyway.

This isn’t a cautious voice on principle. Pachocki simultaneously claims that the current pace could be sustained until the moment when systems begin to meaningfully improve themselves, and that OpenAI is organizing its research around that possibility. That’s a much stronger thesis than simple “let’s be careful.”

What We Don’t Know About Him

Questions about his salary often come up with his name. I won’t give a number, because there isn’t one in any credible source. OpenAI doesn’t publish salaries, and the figures circulating online are estimates based on market ranges, not documents. I’d rather write that I don’t know than repeat someone else’s estimate as fact.

He also says little publicly about himself. He gives few interviews, doesn’t maintain a prominent social media presence, and what you can read about him relates almost exclusively to work.

Wife, Family, Net Worth: What I Won’t Write

People also search for his private life: wife, family, parents, net worth, date of birth, even IQ. You won’t find that here, and I want to explain why, because it’s not an oversight.

Pachocki doesn’t talk about his private life publicly. There’s no interview where he discusses family, no official statement with his birth date, no document with his net worth. The numbers that circulate online with such names are created by multiplying salary ranges by years worked and presenting the result as fact. With IQ it’s even worse, because such numbers attached to public figures are usually invented from scratch.

My rule is simple: if something isn’t in a credible source, I write that I don’t know it, instead of filling the gap with a plausible sentence. With someone who doesn’t talk about himself, that’s the only honest approach. If he ever speaks about it publicly, I’ll update this article.

What we know about him concerns his work, and there’s a lot to write about that: competitions, doctorate, leading science at OpenAI, and what he announced in September 2026. The rest isn’t secret, it’s just not public.

Where This Information Comes From

I verified the dates, medals, and positions on September 11, 2026 in his biography and in OpenAI’s announcements about the change in the Chief Scientist position. I’m quoting the essay content from the version published on OpenAI’s website on September 6, 2026.

I wrote more extensively about the essay itself and what it means for the pace of work on models separately in the article about the call to slow down scaling. I describe the models in question as they develop in the article about GPT-6 Astra.

Photo: Jakub Pachocki at ICPC finals in 2012, photo by ICPCNews, license CC BY 2.0, via Wikimedia Commons.

Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →

Reader Questions

Who is Jakub Pachocki?

Jakub Pachocki is a computer scientist from Gdańsk, Poland, born in 1991. Since May 2024, he has served as Chief Scientist at OpenAI - the company behind ChatGPT. Previously, he led the work on GPT-4. Before entering AI, he was one of the world’s best competitive programmers - he won Google Code Jam in 2012 and won ICPC gold with the second-best result in the world. He earned his PhD from Carnegie Mellon University and joined OpenAI in 2017.

What does Jakub Pachocki do at OpenAI?

As OpenAI’s Chief Scientist, Jakub Pachocki is responsible for the direction of scientific research at the company. GPT-4 was created under his leadership. In September 2026, he published a widely discussed essay “An Alien Mind,” in which he stated that no laboratory - including OpenAI - controls its models well enough to responsibly maintain maximum scaling speed. He advocates for voluntary slowdowns in the industry until shared safety thresholds are established.

How much does Jakub Pachocki earn at OpenAI?

There is no specific data about Jakub Pachocki’s salary in any publicly available source. OpenAI does not disclose employee compensation. Numbers appearing online are market estimates, not data from documents. You can generally check the ranges for a Chief Scientist position at tech companies of this size, but without official data, any specific figure would be speculation.

Kinetyra