Swedish researchers create an ‘AI scientist’ that designs and runs its own experiments

Swedes created an AI system that can carry out the full scientific process from hypothesis to results interpretation with minimal human input.

Researchers in Sweden have developed an AI scientist capable of generating scientific hypotheses, designing experiments and interpreting results.

The team combined multiple Large Language Models (LLMs), large databases and physical laboratory robotics into a closed-loop system that can run the full cycle of scientific discovery.

“It is too much information for a human to analyse, but our AI scientist could identify promising biological questions, recommend experiments to test them, evaluate experimental outcomes and iteratively refine its understanding based on new evidence,” said Ievgeniia Tiukova, a researcher at Chalmers University of Technology, and one of the authors of the new study that examined the AI scientist.

The AI scientist analysed a database of approximately 60,000 phenotypical, physiological and metabolic relations for S. cerevisiae — a species of yeast also called brewer’s yeast or baker’s yeast. Then the system formed nearly 2,000 testable predictions about how nutrients impact cell growth and stress resistance.

Then these hypotheses were evaluated, selected comparison controls and converted the ideas into step-by-step, machine-readable instructions for lab equipment.

Copyright for this text belongs to Euronews.