The UK's Advanced Research and Invention Agency (ARIA) is funding 12 projects to develop AI scientists that can autonomously design and run experiments. Selected from 245 proposals, these teams will receive around £500,000 each over nine months to prove their systems can generate novel scientific findings.

ARIA defines an AI scientist as a system that manages the full workflow: from hypothesis generation to experiment execution and analysis. The agency doubled its initial funding commitment due to the high volume of quality submissions, with teams from the UK, US, and Europe participating. This approach aims to map the cutting edge of automated science.

Winning projects include Lila Sciences' AI NanoScientist for quantum dot research and a University of Liverpool robot chemist using a vision language model. London-based Humanis AI is developing ThetaWorld to study battery chemistry. ARIA's Ant Rowstron views this as a strategic experiment to gauge the pace of change before committing to larger, longer-term projects.