Jakub Pachocki

Overview

Jakub Pachocki, born in 1991 in Gdańsk, Poland, is a computer scientist and the chief scientist at OpenAI.^[1] He joined the company in 2017, became its research director in 2021, and led the development of GPT-4 and OpenAI Five.^[1] Sam Altman has called him "easily one of the greatest minds of our generation".^[1]

Education and early research

Pachocki studied computer science at the University of Warsaw and earned a PhD at Carnegie Mellon University under Gary Miller. He followed that with postdoctoral work at Harvard and at the Simons Institute for the Theory of Computing.^[1]

He was also a competitive programmer of high standing: a silver medal at the 2009 International Olympiad in Informatics, a team gold medal and second place at the 2012 ICPC World Finals, and the Google Code Jam championship in 2012.^[1] Contests of this kind reward exact reasoning and efficient algorithms under a strict time limit, which is close to the kind of work that goes into building and debugging large training systems.

Research at OpenAI

Pachocki joined OpenAI in 2017.^[1] The company's own biography says he has led transformative research initiatives there since 2017, and that he previously served as Director of Research, spearheading the development of GPT-4 and OpenAI Five as well as fundamental research in large-scale reinforcement learning and deep learning optimisation.^[5] The same biography credits him with being instrumental in refocusing the company's vision towards scaling deep learning systems.^[5] In 2021 he became research director.^[1]

"Scaling deep learning systems" names a bet that has shaped much of the field: instead of hand-designing many separate components, build one large model and make it bigger — more parameters, more training data, more computation — and let capability on many tasks improve together. Research in this style is largely about making very large training runs stable and efficient, which is where the optimisation and reinforcement learning background comes in. GPT-4 and OpenAI Five, the two projects the company lists under his leadership, are both products of that approach.^[5]

Chief scientist

During the November 2023 board crisis at OpenAI, Pachocki briefly left the company and was to join the team Sam Altman was planning at Microsoft.^[2] On 14 May 2024, OpenAI announced in a message from Altman that Pachocki would become its chief scientist.^[5] He succeeded Ilya Sutskever in the role.^[1][4]

Safety and public writing

Around the release of GPT-6 in 2026, Pachocki said that preventing unintended harm from AI was becoming increasingly difficult, and that this difficulty could become a bottleneck on AI progress.^[3] In September 2026 he published an essay titled "An Alien Mind", calling for shared safety regulation and for international cooperation on AI alignment — the problem of making a system's behaviour match human intent.^[1]

In a post on X quoted by TechCrunch, he said that OpenAI had worked to preserve and use chain-of-thought monitoring since its very first reasoning models, and that this is a core goal of its current research programme.^[6] Chain-of-thought monitoring means reading the intermediate reasoning a model writes out, rather than only its final answer, so that unwanted lines of thought can be spotted before they turn into unwanted actions.

His path — a contest mathematician who moved into large-scale deep learning, then into a senior scientific role concerned with safety — is a compact picture of how AI research changed: the same people who build the biggest models are now also the ones asked to explain how those models should be checked.