An artificial intelligence platform known as BacterAI, designed by a research team led by a University of Michigan professor, has demonstrated its ability to perform a staggering number of autonomous scientific experiments, up to 10,000 per day. The breakthrough application of AI could pave the way for rapid advances in a variety of fields, including medicine, agriculture, and environmental science.
The results of the study have been published Nature Microbiology.
Decoding Microbial Metabolism with BacterAI
BacterAI was developed to map the metabolism of two microbes associated with oral health without any baseline information. The complex metabolic processes of bacteria involve the use of a specific combination of the 20 amino acids necessary for life. The aim of the study was to identify the exact amino acids that beneficial oral microbes need to promote their growth.
“We know next to nothing about most of the bacteria that affect our health. Understanding how bacteria grow is the first step toward reengineering our microbiome,” said Paul Jensen, a UM assistant professor of biomedical engineering who was at the University of Illinois when the project began.
A difficult task simplified with AI
Deciphering the preferred combination of amino acids for bacteria is a daunting task with over a million possible combinations. However, BacterAI was able to successfully determine the amino acid requirements for the growth of both Streptococcus gordonii and: Streptococcus sanguinis.
BacterAI’s approach involved testing hundreds of amino acid combinations per day, focusing it and changing the combinations daily based on the results of the previous day’s experiments. Within nine days, it achieved 90% accuracy in its predictions.
AI training through trial and error
Unlike traditional methods that use labeled datasets to train machine learning models, BacterAI generates its own dataset through an iterative process of experimenting, analyzing results, and predicting future outcomes. This method enabled him to decipher the feeding rules of bacteria in less than 4,000 experiments.
“We wanted our AI agent to take steps and fall, come up with ideas and make mistakes. Every day it gets a little better, a little smarter,” Jensen said, highlighting the parallels between BacterAI and a child’s learning process.
The future of AI in research
Given that approximately 90% of bacteria have little or no research, conventional methods are a significant bottleneck in terms of time and resources required. BacterAI’s ability to run automated experiments can dramatically accelerate discoveries. In one day, the team managed to perform up to 10,000 experiments.
However, potential applications of BacterAI go beyond microbiology. Researchers in any field can pose questions as puzzles for artificial intelligence to solve through this kind of trial-and-error process.
“With the recent explosion of mainstream artificial intelligence over the past few months, many people are uncertain about what it will bring in the future, positive and negative,” said Adam Dama, a former engineer at Jensen’s lab and lead author of the study. . “But it’s very clear to me that focused applications of AI like our project will accelerate everyday research.”