In a groundbreaking development reported in Nature Microbiology, a team led by a professor at the University of Michigan has harnessed the power of artificial intelligence (AI) to enable robots to conduct up to 10,000 autonomous scientific experiments per day. This cutting-edge technology, known as BacterAI, holds the potential to accelerate the pace of discovery in fields ranging from medicine and agriculture to environmental science.

The researchers, led by Professor Paul Jensen, aimed to understand the metabolism of two microbes associated with oral health, despite having no initial baseline information. Each bacterial species requires specific nutrients to thrive, but determining the precise combination of amino acids they need can be challenging. With 20 amino acids yielding over a million possible combinations, BacterAI stepped in to uncover the amino acid requirements for the growth of Streptococcus gordonii and Streptococcus sanguinis.

Unlike traditional approaches that rely on labeled data sets for machine learning, BacterAI creates its own data set through a series of experiments. By analyzing the results of previous trials, the AI system generates predictions for new experiments that will yield the most informative outcomes. With this iterative process, BacterAI tested hundreds of amino acid combinations daily, adjusting its focus based on previous results. Remarkably, within just nine days, the system achieved 90% accuracy in its predictions.

The significance of BacterAI extends beyond microbiology, as researchers in various disciplines can utilize this AI-driven approach to pose questions and solve puzzles through trial and error. The system’s ability to autonomously generate hypotheses and learn from mistakes resembles the iterative process of a child learning to walk.

Approximately 90% of bacteria remain largely unexplored, and traditional research methods require significant time and resources to uncover even basic scientific information about them. By automating experimentation, BacterAI and similar systems have the potential to accelerate discoveries. The research team conducted up to 10,000 experiments in a single day, highlighting the remarkable speed and efficiency enabled by AI.

As AI continues to gain prominence, it is crucial to focus on specific applications that can drive positive impacts. Adam Dama, the lead author of the study, emphasizes that targeted AI applications, like their project, will expedite everyday research. With AI-powered autonomous experimentation, scientists can unlock new frontiers of knowledge and propel advancements across a wide range of scientific domains.

By Impact Lab