Scientists from the University of Toronto have used artificial intelligence to search for extraterrestrial life. They trained the program to look for signals in radio telescope data that could not be produced by natural astrophysical processes, reports Nature Astronomy. As a result, they recorded 8 radio signals that could potentially come from intelligent beings.
The previously studied dataset was loaded into the artificial intelligence system, and it identified eight signals of potential interest. The classical algorithm failed to identify them.
“We have developed a machine learning system that allows us to search for potential signals of highly advanced technological civilizations in radio waves coming from distant star systems. We were able to single out eight unique extraterrestrial signals at once, which were not observed in the past,” said the astronomers.
At the end of July 2015, Russian entrepreneur Yuri Milner and the late British astrophysicist Stephen Hawking launched the Breakthrough Listen project as part of the Breakthrough Initiatives program, aimed at searching for signals from extraterrestrial civilizations. Within its framework, Milner allocated $100 million to create a network of radio telescopes aimed at searching for so-called “technosignatures”, artificial cosmic signals.
Technosignatures are hypothetical impulses that indicate the presence of an artificial source and, as a result, the existence of a developed society capable of using technology for communication.
To search for “technosignatures”, a group of astronomers led by Breakthrough Initiatives executive director Peter Vorden have developed a machine learning system capable of “sifting” the large amount of data that is collected by the Breakthrough Listen observatories. The work of this algorithm was tested by scientists on data collected since 2016 during observations of the nearest 820 stars using the GBT radio telescope.
Search for extraterrestrial signals
Radio telescopes capture vast amounts of data. However, they contain a lot of interference from sources such as phones, Wi-Fi, and satellites. Therefore, scientists compare the search for extraterrestrial impulses with the search for a needle in a haystack.
The new search algorithm was able to quickly distinguish real radio signals from false positives. During the experiment, over 150 terabytes of data (480 hours of observations) were transferred to the system from the Green Bank Telescope in West Virginia. Artificial intelligence has identified 20,515 signals of interest to researchers. Scientists had to analyze them manually. As a result, they identified eight signals that “had the characteristics of technosignatures and could not be attributed to radio interference.”
These bursts of radio waves were generated by five nearby stars, HIP 56802, HIP 118212, HIP 62207, HIP 54677 and HIP 13402. The last three luminaries, as scientists note, gave rise to two different bursts of radio waves at once, claiming the status of signals from extraterrestrial civilizations.
All of these five stars, as the scientists note, are located at a relatively small distance from Earth, from 30 to 90 light years, which makes them especially interesting for further study. So far, the efforts of astronomers have not been successful – repeated observations of these luminaries using both the GBT and other telescopes have not led to the re-fixation of those “technosignatures” that were found by the AI system in the data of the Breakthrough Listen project.
The authors of the study said they intend to monitor such signals which could also be rare radio interference but nevertheless, the experiment on the introduction of artificial intelligence was considered successful. With further advances in technology, the system will be able to more effectively separate real signals from interference.
In the course of further research we will receive new evidence of the presence of extraterrestrial civilizations, but this may take some years.