.png)
April 23, 2025
The search for life outside our solar system is one of the greatest challenges of modern astronomy. The sheer number of potential candidates and the difficulty of identifying Earth-like planets make the search a complex task. A new AI model, developed by researchers at the University of Bern, could now significantly accelerate this search. The model identified 44 star systems that could potentially harbor habitable planets.
The identification of exoplanets, planets outside our solar system, is extremely difficult. Their small size and lack of their own light make them almost invisible compared to the brightly shining stars. To make the search more efficient, researchers are increasingly relying on artificial intelligence. The team from the University of Bern trained an AI model with synthetic data based on the "Bern Model of Planet Formation and Evolution." This model simulates the formation of planets from gas and dust disks around young stars, taking into account a variety of physical factors.
By training with this simulated data, the AI learned to recognize patterns in planetary systems and predict whether a system could harbor Earth-like planets. Particularly relevant for the prediction were the properties of the innermost known planets in a system, especially their mass and orbital period. These parameters provide clues to the possible existence of further planets in the habitable zone, the region around a star where water can exist in liquid form – a basic prerequisite for life as we know it.
The AI model was subsequently tested with data from nearly 1,600 known star systems, each containing at least one confirmed planet and a star of type G, K, or M – i.e., sun-like or cooler stars. The AI marked 44 of these systems as particularly promising for the existence of Earth-like planets in the habitable zone. The selection was based on specific factors, such as the mass and orbital period of inner planets, which suggest the possible presence of other, potentially habitable planets.
Despite the high prediction accuracy of up to 99 percent in the test with simulated data, the model also has limitations. For example, it could not fully reproduce some known relationships, such as the co-occurrence of super-Earths and cold gas giants. Also, the positions of the simulated planets sometimes deviated from the real data, which can influence the prediction accuracy regarding habitability. The distance of a planet from its star is a decisive factor for the existence of liquid water and thus for the possibility of life.
Despite these limitations, the study demonstrates the potential of AI in the search for habitable planets. The results offer valuable clues for future astronomical investigations and enable a more targeted use of telescopes. Instead of randomly searching the sky, researchers can now focus on the most promising candidates, thus increasing the chances of discovering life outside our solar system.
Bibliographie: - https://t3n.de/news/ki-44-sternensysteme-leben-1684194/ - https://x.com/t3n/status/1914728222148747312 - https://t3n.de/tag/future-science/ - https://t3n.de/tag/kuenstliche-intelligenz/ - https://newstral.com/de/article/de/1265867177/ki-entdeckt-44-sternensysteme-mit-m%C3%B6glichem-leben - https://t3n.de/ - https://t3n.de/news/ - https://t3n.de/tag/innovation/