OpenAI is throwing everything into building a fully automated researcher
AI Summary: Researchers at OpenAI are advancing the capabilities of large language models (LLMs) to tackle long-running scientific research tasks, building on the success of coding agents like Codex. The development of reasoning models has enhanced the ability of these systems to work independently for extended periods, enabling them to manage complex problems by breaking them into subtasks. Recent applications of GPT-5 have led to new solutions for unsolved mathematical problems and breakthroughs in biology, chemistry, and physics. OpenAI aims to apply the problem-solving capabilities demonstrated in coding to broader scientific challenges, prioritizing real-world relevance over theoretical achievements.