OpenAI Targets 2028 for ‘Legitimate AI Researcher’ Capable of Autonomous Problem Solving

OpenAI is significantly refocusing its research efforts toward building a fully automated AI researcher, aiming for a functional prototype by 2028. This ambitious project represents a shift in the company’s strategy, consolidating various research strands like reasoning models, agents, and interpretability under a single, overarching goal.

The Road to Autonomous Research

The initiative will unfold in stages. By September 2026, OpenAI plans to launch an “autonomous AI research intern” capable of tackling a limited set of specific research problems independently [TechCrunch]. This intern will serve as a precursor to the fully automated, multi-agent system slated for release in 2028 [MIT Technology Review].

Potential Applications Span Multiple Disciplines

OpenAI envisions this AI researcher tackling problems currently too complex for human researchers. Potential applications span a wide range of fields, including mathematics, physics, biology, chemistry, business, and policy [MIT Technology Review]. The system is designed to accept problems formulated in text, code, or visual representations, broadening its potential scope.

Jakub Pachocki Leads the Charge

A significant portion of this strategic decision rests with Jakub Pachocki, OpenAI’s chief scientist, alongside chief research officer Mark Chen [MIT Technology Review]. Pachocki described the AI researcher as a “system capable of autonomously delivering on larger research projects” [TechCrunch].

Scaling Compute and Advancing Deep Learning

OpenAI’s strategy relies on two key pillars: continued algorithmic innovation and dramatically increasing “test time compute”—the amount of time models dedicate to analyzing problems [TechCrunch]. Current models can address tasks with a roughly five-hour computational horizon and achieve performance comparable to top human competitors in fields like the International Mathematical Olympiad. OpenAI anticipates a rapid expansion of this horizon through increased computational resources.

Competition and the Future of AI

OpenAI’s move comes amid increasing competition from companies like Anthropic and Google DeepMind. The company’s future direction will have significant implications for the broader AI landscape [MIT Technology Review]. Pachocki believes deep learning systems could be less than a decade away from achieving superintelligence – systems smarter than humans across a wide range of tasks [TechCrunch].