An exclusive conversation with OpenAI’s chief scientist Jakub Pachocki about his firm's new grand challenge and the future of AI.

Technology Review, published by MIT, is a leading source of news, analysis, and commentary on emerging technologies, scientific advancements, and their societal impacts. Covering topics such as artificial intelligence, biotechnology, renewable energy, and more, Technology Review provides  reporting and  articles on the intersection of technology and society. Readers can stay informed about the latest breakthroughs and trends shaping the future of technology.

MIT Technology Review

OpenAI has announced a new strategic direction: building a fully automated AI researcher. Chief scientist Jakub Pachocki outlines a two-phase plan — an autonomous AI research intern by September 2025, followed by a full multi-agent research system by 2028. The system would tackle complex problems in math, biology, chemistry, and policy autonomously. Codex, OpenAI's agent-based coding tool, is cited as an early prototype. Pachocki discusses how improvements in reasoning models and long-context handling are enabling longer autonomous operation. He also addresses safety concerns, noting that chain-of-thought monitoring is currently the best tool for oversight, while acknowledging that fully controlling such systems remains unsolved and will require government involvement.

OpenAI is throwing everything into building a fully automated researcher