According to a study published in the journal Nature Microbiology, researchers from the University of Michigan have developed a novel AI platform, dubbed BacterAI, that can help robots conduct approximately 10,000 automated scientific experiments per day.

The researchers revealed that the AI model helped map the metabolism of two microbes, namely Streptococcus gordonii and Streptococcus sanguinis. Both of the microbial strains of Streptococcus are associated with oral health, and the researchers had no baseline information to start with.

They explained that bacteria require a combination of 20 amino acids to support life, but each species requires specific nutrients to grow. However, mapping the combinations of amino acids that bacteria like is tricky due to the millions of possible combinations. In the study, to determine the right formula for each species, BacterAI tested hundreds of combinations of amino acids per day, honing its focus and changing combinations each morning based on the previous days results.

The researchers revealed that within nine days, the AI model produced correct results with an accuracy of 90%. They explained that BacterAI develops its data set through a series of trials, unlike traditional techniques that input labeled data sets into a machine-learning model.

They also noted that the model can predict the correct test sequence that will provide the most information by examining the outcomes of past ones. As a result, in less than 4,000 attempts, the model can predict the combination of amino acids required to support bacterial growth.

(Source: https://www.thehansindia.com/hans/young-hans/novel-ai-enabled-robot-to-conduct-automated-scientific-experiments-796141)