OpenAI is throwing everything into building a fully automated researcher
AI Summary
OpenAI has announced a major strategic refocus, designating the development of a fully automated AI researcher as its 'North Star' for the coming years, according to an exclusive interview published by MIT Technology Review on March 20, 2026. The company has set a concrete near-term milestone of building an 'autonomous AI research intern' — a system capable of independently handling research tasks that would take a human several days — by September 2026, with a fully automated multi-agent research system targeted for debut in 2028. OpenAI Chief Scientist Jakub Pachocki, who played key roles in the development of GPT-4 in 2023 and reasoning models in 2024, described the vision as eventually housing 'a whole research lab in a data center,' capable of tackling complex problems in mathematics, physics, biology, chemistry, and policy. The company points to its existing agent-based coding tool Codex, powered by GPT-5, as an early precursor to this vision, noting that most of OpenAI's technical staff now use it and that it has already helped researchers discover new solutions to unsolved math problems. On safety, Pachocki acknowledged 'serious unanswered questions,' citing risks including system misalignment, hacking, and misunderstood instructions, and said OpenAI is currently relying on chain-of-thought monitoring — a technique where models log notes about their actions in real time — as its primary oversight mechanism, with new details on its use with Codex published the same day. Pachocki also stated that even by 2028, he does not expect AI systems to match human intelligence across all dimensions, framing the goal instead around building 'economically transformative technology' rather than AGI.
Why it matters
OpenAI's articulation of a concrete multi-year roadmap — with a September 2026 intern milestone and a 2028 full system target — signals an intensifying race among top AI labs, including Anthropic and Google DeepMind, to capture the high-value scientific and enterprise research automation market, a segment with potentially significant revenue implications across pharmaceuticals, materials science, and software development. The explicit acknowledgment that OpenAI recently signed a deal with the Pentagon following Anthropic's public dispute with the U.S. military highlights the growing role of government contracts as a competitive differentiator and revenue stream in the AI sector. For investors tracking the AI infrastructure and enterprise AI space, the emphasis on long-running autonomous agents requiring sustained compute in large data centers reinforces demand signals for AI hardware and cloud infrastructure providers.
Scoring rationale
This article directly covers OpenAI's major strategic research pivot toward fully autonomous AI agents and researchers, with specific timelines and competitive implications that are highly material to AI market dynamics and OpenAI's valuation trajectory.
Impacted tickers
This summary was generated by AI from the original article published by MIT Technology Review AI. AIMarketWire does not provide trading advice. Always refer to the original source for complete reporting.