Stories
AI Scientist Runs Own Biology Lab, Makes Discoveries
- •Researchers developed an AI scientist that can generate biological hypotheses, design experiments, direct laboratory automation, analyze results, and refine experiments.
- •The system, developed by Chalmers University of Technology, University of Gothenburg, and University of Cambridge, was tested on Saccharomyces cerevisiae (brewer’s yeast).
- •Combining large language models, automated reasoning, and laboratory robotics, the AI identifies biological questions, generates hypotheses, and converts experimental plans into instructions for automated equipment.
- •The AI generated nearly 2,000 testable predictions on how nutrients affect yeast growth and stress resistance, identifying previously underexplored biological interactions.
- •While not fully autonomous, the AI acts as a research partner, accelerating biological investigations by automating hypothesis generation and experimental testing, with human oversight for broader research boundaries and interpretation.
Key Questions Answered
The system combines large language models, automated reasoning, and laboratory robotics into a closed-loop lab. It uses existing knowledge to generate hypotheses, converts experimental plans into instructions for automated equipment, analyzes the results, and refines future experiments.
The AI scientist system was tested on Saccharomyces cerevisiae, which is commonly known as brewer's or baker's yeast. Researchers tested it using a database containing about 60,000 phenotypic, physiological, and metabolic relationships for the organism.
The new system builds on earlier work by Professor Ross King and his colleagues, who previously developed Adam as the first general-purpose robot scientist and Eve, a robot designed specifically for drug discovery.
