Jakub Pachocki. Chief Scientist at OpenAI, where he leads core research direction and has been instrumental in the development of frontier language models. Pachocki has worked on scaling, alignment, and safety research, and speaks publicly on the tension between capability acceleration and the need for defensive AI systems.
Operating themes
- Defensive AI rationale How building more capable aligned systems is the practical response to threats posed by advanced AI.
- Capability and safety as linked The argument that accelerating capability research is not in tension with safety when the goal is building protective systems.
- Frontier model development The internal dynamics of training, evaluating, and deploying models at the research frontier.