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

[Autonomous AI Scientist Created by Swedish Research Team That Designs and Executes Experiments]

Researchers in Sweden have created an AI scientist that can generate hypotheses, design experiments, and interpret results.

The team integrated multiple large language models, extensive databases, and physical laboratory robotics into a closed‑loop system capable of completing the entire scientific discovery pipeline.

Ievgeniia Tiukova, a researcher at Chalmers University of Technology and co‑author of the study, explained that while the sheer volume of available information exceeds human analysis capability, the AI scientist can pinpoint promising biological questions, propose experiments to test them, assess the outcomes, and continuously refine its understanding using new data.

By querying a repository of roughly 60,000 phenotypic, physiological, and metabolic relationships for Saccharomyces cerevisiae—commonly known as brewers’ or bakers’ yeast—the system produced almost two thousand testable predictions on how nutrients affect cell growth and stress resistance.

The generated hypotheses were evaluated against a suite of comparative controls, then transformed into sequential, machine‑readable commands for laboratory equipment.

After experiment completion, the AI analyzed the resulting data, confirmed or rejected the original predictions, and automatically adjusted or discarded hypothesis statements for upcoming cycles.

“Rather than serving solely as decision‑support tools, the AI scientist proactively generates novel scientific knowledge,” said Tiukova.

The authors emphasized that the future of scientific inquiry depends heavily on increasing automation; advances in robotics, artificial intelligence, and high‑throughput technology are reshaping research workflows.

Ross King, senior author at the University of Gothenburg, added that ‘AI scientists will collaborate with human researchers to accelerate breakthroughs across biology, medicine, and biotechnology.’

“Future generations of autonomous laboratories will become increasingly capable of working hand‑in‑hand with humans and will serve as valuable partners in addressing some of the most demanding questions in biology and medicine,” King concluded.

Humans will still be essential

Researchers said such AI solutions could minimise bias, accelerate research cycles, and reduce variability stemming from human error, incomplete record‑keeping, and fluctuating laboratory conditions, yet human involvement remains indispensable in directing priority areas, interpreting broader contexts, planning comprehensive research programmes, and upholding ethical compliance.

However, the authors stressed that humans will still be key in scientific research.

“Humans are fundamental for defining priority areas, interpreting broader context, designing research programmes, and upholding ethical responsibilities,” King affirmed.

“Emerging autonomous laboratories are poised to become increasingly adept at partnering with scholars, furnishing dynamic synergy to tackle the most formidable challenges in biology and medicine,” he added.

International Guidelines Emerge

As AI use in research expands, international bodies are crafting governance frameworks. The World Health Organization recently issued a report containing recommendations for researchers, ethics committees, regulators, funders, and policymakers to guide responsible development and deployment of AI in health research.

“Artificial intelligence presents unprecedented opportunities to accelerate health research and enhance public well‑being,” said Meg Doherty, director of WHO’s Department of Science for Health.

“Innovation must be steered by robust ethical safeguards that protect human dignity, rights, and equity,” she continued.

Source link