An artistic illustration of the virus. (Image credit: THOM LEACH/SCIENCE PHOTO LIBRARY via Getty Images)
Google's new artificial intelligence (AI) tool has solved a problem that took scientists a decade to solve in just two days.
Jose Penades and his team at Imperial College London have spent 10 years studying the mechanisms by which some superbugs become resistant to antibiotics – a growing threat that claims millions of lives every year.
However, when the researchers asked Google's “fellow scientist” – an AI tool designed to assist researchers – this question in a short query, the AI's response matched their as-yet-unpublished results within just two days.
Surprised, Penades messaged Google to see if they had access to his research. The company said no. The researchers posted their findings on February 19 on the preprint server bioRxiv, meaning their work has not yet been peer-reviewed.
“Our results show that AI can synthesize all the available data and guide us to the most important questions and experimental designs,” said co-author Thiago Dias da Costa, a lecturer in bacterial pathogenesis at Imperial College London. “If the system works as well as we expect, it could be a game-changer, eliminating dead ends and allowing us to make progress at incredible speed.”
Using AI to Fight Superbugs
Antimicrobial resistance (AMR) occurs when infectious microbes — such as bacteria, viruses, fungi and parasites — develop resistance to antibiotics, rendering essential drugs ineffective. Dubbed a “silent pandemic,” the problem is one of the greatest threats to human health, with overuse and misuse of antibiotics in both medicine and agriculture fueling its spread.
According to a 2019 report from the Centers for Disease Control and Prevention (CDC), drug-resistant bacteria caused at least 1.27 million deaths worldwide that year. About 35,000 of those cases were reported in the U.S. alone, meaning U.S. deaths from the problem have increased by 52% since the CDC’s last AMR report in 2013.
To study this problem, Penades and his team began looking for ways in which one type of superbug—a family of bacteria-infecting viruses known as capsid-forming phage-inducible chromosomal islands (cf-PICIs)—evolves the ability to infect different types of bacteria.
Scientists hypothesized that these viruses do this by borrowing the tails used to inject the viral genome into the host bacterial cell from various viruses that infect bacteria. Experiments confirmed their hypothesis, revealing a new mechanism for horizontal gene transfer that the scientific community had not previously known about.
Before any of the team shared their findings publicly, the researchers asked the same question to Google's AI co-scientist tool. Two days later, the AI came up with ideas, one of which matched their assessment.
“This effectively shows that the algorithm was able to process the available data, analyse possibilities, ask questions, design experiments and come up with the same hypothesis that we arrived at through years of hard scientific research, but in a fraction of the time,” Penades, a professor of microbiology at Imperial College London, said in a statement.
The researchers stressed that using AI from the start would not eliminate the need for experiments, but would help them come up with hypotheses much earlier, saving years of work.
Despite promising results and other advances, the use of AI in science remains a subject of debate. For example, more and more studies using AI are found to be irreproducible or even outright fraudulent. To minimize these problems and maximize the usefulness that
Sourse: www.livescience.com
