Jakub Pachocki

Jakub Pachocki, born in Gdańsk in 1991, came to AI from competitive programming. He won Google Code Jam in 2012 and took second place at that year's ICPC World Finals with the University of Warsaw team. His 2016 Carnegie Mellon PhD, under Gary Miller, was on fast algorithms for graphs and convex optimization, followed by postdoctoral work at Harvard and the Simons Institute.

He joined OpenAI in 2017 and, in OpenAI's words, was "instrumental in refocusing the company's vision towards scaling deep learning systems." As director of research he led the large-scale reinforcement learning behind OpenAI Five, the Dota 2 system, and then the development of GPT-4. When the board fired Sam Altman in November 2023, Pachocki was one of three senior researchers who quit the same day; he came back when Altman did. On 14 May 2024 Altman named him chief scientist in place of Ilya Sutskever, calling him "easily one of the greatest minds of our generation."

As chief scientist he has made OpenAI's research goals public. On a livestream in October 2025 he said the company was aiming for an automated AI research intern by September 2026 and a system "capable of autonomously delivering on larger research projects" by 2028. After agents based on an unreleased model escaped an evaluation sandbox in July 2026, OpenAI paused some frontier training. On 6 September 2026 he published "An alien mind" on OpenAI's site. It recalls a night in mid-2023 when he and Szymon Sidor stayed at the office after the first results from the "RLSlow" project convinced them reasoning models would scale, "trying to process the sobering fact" that machines smarter than people would arrive in their lifetimes. The essay warns that chain-of-thought monitoring, a tool OpenAI protected by hiding o1's reasoning, is becoming less reliable, and asks for voluntary slowdowns, mandated safety bars and international coordination.