Astronomers have a new tool in their arsenal to help them hunt down the most enormous explosions in the universe. An international team led by Northwestern University has successfully created an artificial intelligence (AI) tool that can detect, identify, and classify supernovas—a fully automated system that cuts out the human middleman and could dramatically speed up the discovery process. This machine—which is named Bright Transient Survey Bot, and no, the researchers weren’t fans of the wildly popular South Korean boy band—recently discovered a supernova dubbed SN2023tyk using data from a robotic telescope called the Zwicky Transient Facility (ZTF). The BTSbot then automatically sent that candidate to another robotic telescope known as the Spectral Energy Distribution Machine (SEDM) at Caltech in California for further observation to collect its spectrum, an essential piece of evidence to confirm if a star is exploding.
The SEDM then uses a machine-learning algorithm to classify the object based on its brightness, which usually indicates what type of stellar explosion it is, such as a supernova or a flaring star. If the classification passes muster, the BTSbot will notify astronomers of its findings, and the results can be published in scientific journals. The whole process only takes days, as opposed to the usual weeks or months that can pass between when a candidate is first observed and when it is confirmed as an actual supernova.
To develop the BTSbot, a machine-learning algorithm was trained to recognize these cosmic explosions by looking at more than 1.4 million images from 16,000 sources—including previously confirmed supernovae, flaring stars, and temporarily flaring galaxies. Once programmed to spot these events, the BTSbot could quickly and efficiently shift through data from ZTF to identify a potential candidate. Then, it would send that information to the SEDM to request a closer look at the source and confirm whether it was an exploding star.
After the BTSbot successfully detected SN2023tyk, it publicly shared its discovery with the astronomical community on Oct. 7, four days after ZTF initially spotted the event. This success marks the first time that an AI has been able to observe, identify, confirm, and classify a supernova in real time. Ultimately, the BTSbot will free up astronomers’ time to spend on other research, such as developing fresh hypotheses about the origin of these spectacular stellar explosions.
The team behind the BTSbot consists of a multi-disciplinary group of astronomers and scientists from Caltech, the University of Minnesota, Liverpool John Moores University in England, and Stockholm University in Sweden. Their work was published in the journal Science Advances. Gizmodo’s take: This is just the latest example of how AI is starting to change how we live, work, and play. Sign up for our weekly newsletter to stay on top of the latest tech. You can also follow us on Facebook, Twitter and Instagram.