MIT tech review: OpenAI is Building an Automated Researcher

4 hours ago

OpenAI is focusing on building a fully automated AI researcher, aiming to tackle large, complex problems independently.
The company plans to debut an 'AI research intern' by September and a full multi-agent research system by 2028.
Jakub Pachocki, OpenAI's chief scientist, leads this initiative, emphasizing models that can work autonomously for extended periods.
OpenAI's Codex is an early version of this vision, already used by technical staff for tasks like coding and document analysis.
The goal is to extend Codex's capabilities beyond coding to broader scientific and real-world problem-solving.
OpenAI acknowledges risks, such as misuse or unintended consequences, and is developing monitoring techniques like chain-of-thought.
The company stresses the need for policy involvement to manage the societal impact of powerful AI systems.
Despite progress, OpenAI remains cautious, deploying powerful models in sandboxes to prevent harm.
The initiative reflects OpenAI's mission to ensure AI benefits humanity, though AGI remains a distant, debated goal.

Hasty Briefsbeta