Two AI-based science assistants succeed with drug-retargeting tasks
Ars Technica AI · John Timmer · 2026-05-19
Google's Co-Scientist and FutureHouse's AI system both demonstrate success at biological drug-retargeting tasks in Nature papers, using agentic architectures to process scientific literature rather than replace researchers.
Appears in
Extraction
Topics: ai-science-assistantsdrug-discoveryagentic-aibiological-research
Claims
- Google's Co-Scientist and FutureHouse's AI system both achieved results on drug-retargeting tasks, with findings published simultaneously in Nature.
- Both systems are agentic, operating in the background by calling out to separate specialized tools.
- Neither system is intended to replace scientists, but to help process the overwhelming volume of scientific literature.
- FutureHouse's system goes beyond Google's by being able to evaluate biological data from specific classes of experiments, not just synthesize literature.
- Microsoft has taken a similar agentic approach to science assistance, while OpenAI instead fine-tuned an LLM directly for biology.
Key quotes
this is not an attempt to replace either scientists or the scientific process. Instead, it's meant to help with what current AIs are best at: chewing through massive amounts of information that humans would struggle to come to grips with.
One, Google's Co-Scientist, is designed as what they term 'scientist in the loop,' meaning researchers are regularly applying their judgments to direct the system.