A scientific paper that answers back: AI turns published studies into working research assistants

Scientific papers usually just sit there — you read one, then spend hours or days trying to make its code and methods work on your own data. A Stanford team led by computer scientist James Zou has built Paper2Agent, a system that turns a paper, its code and its data into a ready-made tool kit an AI assistant can actually use. In tests, the team converted 100 computational biology papers and got 74 of them working as 'paper-agents' — the rest were missing code or data, or had software that would not run. The converted papers answered questions and ran analyses more accurately, and more than twice as fast, as the same AI working from the raw paper and code alone. Zou's team has already made over 100 paper-agents from its own work, and says they have even helped spot new collaborators.

SDG 9 · Industry, innovation and infrastructure — Mechanism: tech innovation