AI aids the search for planets … and life
HEADLINES IN 2023 were often dominated by the way AI is changing our world. And although the use of machine-learning tools in astronomy isn’t new, the practice began to see more attention in 2023 (including a feature in our July issue).
A Jan. 30, 2023, paper in Nature Astronomy showed how scientists searching for alien civilizations used AI to sift through nearly 500 hours of radio signals from over 800 stars. They were looking for patterns that couldn’t be natural, while throwing out interference from human technology. The algorithm pared down nearly 3 million events to just 20,515, which were examined by eye to ultimately identify eight possible technosignatures — signs of a technologically advanced civilization — from five stars. The signals were not seen when these stars were re-observed — so we haven’t found aliens yet. But the researchers noted the technique had fulfilled its purpose by identifying specific signals for follow-up.
An April 21 paper in The Astrophysical Journal showed that machine-learning tools can identify planets forming in the disk of dust and gas around a star. A team led by Jason Terry of the University of Georgia in Athens developed an algorithm to search images for the subtle signs of fledgling planets, which affect the orbit of nearby material and eventually carve out gaps in the disk. Not only did their model “rediscover” known planets, it also flagged a planet around the star HD 142666 that researchers hadn’t spotted. Terry’s team followed up and confirmed a likely forming planet there, demonstrating the model’s potential. “We think there will be an important place for these types of techniques as our datasets get even larger,” Terry said in a press release.
On Sept. 25, a paper in Proceedings of the National Academy of Sciences presented a machine-learning algorithm that could determine whether a sample of material was produced by life or through natural (abiotic) processes. And it could do so with 90 percent accuracy. The technique can be applied with existing technology and used on future space missions — or trace the history of ancient life on Earth.
These are just highlights from a year filled with AI-assisted discoveries. There will doubtless be many more in the years to come.