OpenAI is throwing everything into building a fully automated researcher

March 21, 2026 - 11:17

In an exclusive conversation, OpenAI's Chief Scientist Jakub Pachocki has outlined the company's ambitious new frontier: developing a fully automated AI researcher. This project represents a monumental shift, aiming to create artificial intelligence capable of independently conducting the entire scientific process.

Pachocki described the goal as building an AI that can not only read and comprehend vast scientific literature but also identify gaps in human knowledge, formulate novel hypotheses, design and run experiments through code or simulations, and ultimately interpret the results. This push signifies that OpenAI is channeling significant resources and focus into what it views as the next evolutionary step for artificial general intelligence (AGI).

The implications of such a technology are profound. A reliable automated researcher could dramatically accelerate the pace of discovery across fields like medicine, materials science, and climate research, solving complex problems that currently bottleneck human teams. However, Pachocki acknowledges the immense technical hurdles, including ensuring the AI's reasoning is robust, its findings are reliable, and it can navigate the immense complexity of real-world scientific inquiry.

This initiative moves beyond chatbots and content creators, positioning AI as an active partner in expanding the boundaries of human understanding. While the timeline remains uncertain, OpenAI's commitment signals a future where AI may become the primary engine for groundbreaking discovery.