Researchers in Switzerland have created an artificial intelligence model to detect unknown worlds that could be suitable for life. Their machine learning algorithm has already identified 44 star systems with such planets.
Researchers in Switzerland have created an artificial intelligence model to detect unknown worlds that could be suitable for life. Their machine learning algorithm has already identified 44 star systems with such planets.
This is reported by Futurism, citing a study published in the journal Astronomy and & Astrophysics. It is noted that it is a promising demonstration of how AI can accelerate the search for planets where life may exist.
The AI model did not confirm that Earth-like planets are actually there, but it did encourage astronomers to explore these regions of the galaxy in the future. In simulations, the model achieved an accuracy of 0.99, meaning that 99% of the detected systems have at least one Earth-like planet.
«This is one of the few models in the world with such a level of complexity and depth that it allows for predictive studies like ours. This is a significant step in the search for planets with conditions favorable for life and, ultimately, for the search for life in the Universe,» co-author Dr. Jan Alibert, co-director of the Center for Space and Habitability at the University of Bern, said in a statement.
Exoplanets are difficult to detect because they are tiny compared to stars and produce little of their own light. Scientists have so far confirmed the existence of just over 5,800 planets outside our solar system, and very little is known about them.
This doesn’t provide much material for training pattern-finding algorithms, which require huge amounts of data. Instead, the astronomers loaded their model with synthetic planetary systems created using the Berne Model of Planetary Formation and Evolution, which comprehensively simulates the development of hypothetical planets all the way back to their birth from a protoplanetary disk.
«The Bern model is one of the few models in the world that offers such a wealth of interconnected physical processes and allows for research like the current one», — said Alibert.
During these tests, the AI model showed that the strongest signs of an Earth-like planet can be found in the most distant detectable planet in the system, including its mass and orbital period, the researchers wrote in the study.
The team then applied the machine learning algorithm to a sample of nearly 1,600 systems that have at least one known planet and a G, K, or M-type star—G-type stars are stars similar to the Sun, while the other two classifications describe stars that are smaller and cooler—and found that nearly four dozen of them were likely to host an Earth-like world.
But the model is not foolproof. It doesn’t reproduce certain characteristics of star systems that astronomers have observed, such as the strong correlation between so-called super-Earths and cold Jupiters, which often appear together around stars like the Sun, the authors note. And synthetic planets tend to be closer to their stars than real ones.
However, it still allows astronomers to narrow down the search in the vast universe for planets similar to Earth.
As a reminder, astronomers from the University of Cambridge recently used NASA’s James Webb Space Telescope to detect signs of life in the atmosphere of a planet 124 light-years from Earth. They could indicate that K2-18b is teeming with life and that life is widespread in the galaxy.
Previously, scientists from the Taras Shevchenko National University of Kyiv proposed using asteroids that orbit between Earth, Mars, and Venus to protect astronauts from cosmic radiation during interplanetary travel.
