AI is now transforming scientific research, as one artificial intelligence platform could be capable of ramping up research efforts significantly by conducting as many as 10,000 microbial experiments each day and potentially a million in a year.
BacterAI: AI Scientist
According to Science Daily, an AI platform known as BacterAI was able to map the metabolism of two different microbes linked to oral health. This was done even without any baseline information.
Bacteria are known to consume a mix of 20 amino acids necessary for life. However, each species has specific nutrient requirements for its growth.
A research team from the University of Michigan wanted to know the specific amino acids that beneficial oral microbes need for growth promotion. They published their findings in the Nature Microbiology journal.
According to a statement from the university, Paul Jensen, a biomedical engineering assistant professor who was at the University of Illinois during the start of the project, notes that they barely know anything about health-influencing bacteria. Professor Jensen adds that knowing bacteria's mechanisms of growth is the first step to take when it comes to microbiome reengineering.
It is quite tricky to point out the specific amino acid mix that bacteria like. There are over a million possible mixes that can be inferred from the 20 amino acids' presence or absence. However, BacterAI was able to impressively pinpoint the required amino acids for Streptococcus sanguinis and Streptococcus gordonii growth.
The AI scientist evaluated hundreds of amino acids mixed each day. It honed its focus and altered mixes every morning depending on the results of the previous days. Science Daily reports that, in a span of nine days, BacterAI was able to accurately predict roughly 90% of the time.
Unlike other machine-learning approaches, BacterAI's data set was made through various experiments. By looking into the results of prior experiments, it offers predictions regarding new experiments that may offer the most data. Because of this, BacterAI was able to understand most bacterial feeding rules in less than 4,000 experiments.
ALSO READ: AI Doctor? ChatGPT Nearly Passes US Medical Licensing Exam
10,000 Microbe Experiments Daily, a Million Yearly
There has been minimal research on 90% of bacteria. Moreover, the time and resources required to study them conventionally are quite taxing. But with automated experimentation, which is what BacterAI offers, discoveries can be significantly sped up.
In fact, with BacterAI, the team was able to run up to 10,000 experiments in a single day. Science Daily also reports that the AI could also conducts up to a million microbial experiments a year.
While the specific applications in this case were for microbiology, applications may go beyond this field. Scientists from any field of expertise may set up puzzle questions for the AI to solve.
Adam Dama, the study's lead author and a former engineer from the Jensen Lab, explains that with the recent surge of AI, there seems to be circulating uncertainty about the positives and negatives that the technology may offer in the future. Dama adds, however, that it is quite clear to him that focused AI applications, such as what was demonstrated in their project, will boost everyday scientific research in agriculture, medicine, and environmental science, among others.
RELATED ARTICLE: Artificial Intelligence Pilot? Self-Flying F-16 Craft Takes Off, Combats, and Touches Down All On Its Own
Check out more news and information on Artificial Intelligence in Science Times.
