Forget co-pilot mode — this one wants the whole lab coat. Researchers in Sweden built an “AI scientist” that can generate hypotheses, design experiments, run them on laboratory robots, and interpret the results with minimal human input, Euronews reported Friday.
The closed-loop system stitches together multiple large language models, big databases, and physical robotics. It mined roughly 60,000 phenotypical, physiological, and metabolic relations for baker’s yeast (S. cerevisiae), then formed nearly 2,000 testable predictions about how nutrients affect cell growth and stress resistance — converting winners into machine-readable instructions for lab gear.
“It is too much information for a human to analyse, but our AI scientist could identify promising biological questions… and iteratively refine its understanding based on new evidence,” said Chalmers University researcher Ievgeniia Tiukova. Senior author Ross King of the University of Gothenburg said such systems will collaborate with humans to accelerate biology, medicine, and biotech — while humans still set priorities, context, and ethics.
Automation pitch, meet the reproducibility crisis. The team says AI labs can cut human bias and protocol drift — but WHO is already publishing governance guidance for AI-enabled health research. The robots are learning the scientific method. The referees still wear name badges.
Sources: Euronews
Leave a Reply