An astronomer is employing artificial intelligence to accelerate the discovery of Earth-like exoplanets. This innovative approach aims to identify the stars most likely to host habitable planets, saving astronomers time and effort while increasing the chances of finding a “second Earth” among thousands of planets discovered outside our solar system.
Article contents:
- What is meant by Earth-like planets?
- Artificial intelligence enters the astronomical search line
- A model that learns from data to identify promising stars
- Encouraging preliminary results despite the rarity of terrestrial planets
- Continuing the search: Focusing on younger stars for longer
- A powerful tool despite the challenges
- The first steps in the astrobiology journey
- Frequently Asked Questions
What is meant by Earth-like planets?
When scientists talk about an “Earth-like” planet, they usually mean a planet that is close to Earth in mass and orbits a star within the so-called “habitable zone” of that star. The habitable zone is the range of distances around a star where liquid water could potentially exist on the planet’s surface. However, the presence of a planet in this range does not necessarily mean it teems with life or even possesses oceans and an atmosphere like Earth—it is merely a preliminary condition for the possibility of life as we know it, not a guarantee of it.
AI camera to detect aliens in the sky soon
Artificial intelligence enters the astronomical search line
In this context, astronomer Jeanne Davoult (a French researcher at the German Aerospace Center in Berlin) is leading efforts to use artificial intelligence to speed up the process of finding Earth-like planets. Davoult recently published a scientific paper in the journal »Astronomy & Astrophysics« explaining that she and her colleagues developed an AI model to predict which stars are most likely to possess an Earth-like planet. The main idea is to employ artificial intelligence to avoid prolonged random searching, reduce telescope observation time, and increase the number of potential discoveries.
A model that learns from data to identify promising stars
Davoult and her team relied on a previous study that found certain correlations between the presence of a terrestrial planet in a stellar system and the characteristics of that system. Using the results of that study, they trained a “random forest” algorithm—a type of machine learning algorithm—on data from known planetary systems so that it learns to distinguish between systems likely to contain an Earth-like planet and those unlikely to do so. Practically speaking, this model examines known planetary systems and classifies them into two categories: “a system hosting an Earth-like planet” or “a system lacking an Earth-like planet,” based on the patterns it recognized during the training phase.
Davoult explains the idea by saying: »In planet discovery, we try to identify patterns in datasets that indicate the presence of a planet«. Her team adds in the research paper: »Understanding where Earth-like planets form first, and consequently focusing observations there to avoid random search operations, reduces the average observation time required to discover an Earth-like planet and increases the number of discoveries«.
Encouraging preliminary results despite the rarity of terrestrial planets
So far, despite the discovery of about 6,000 exoplanets in the last thirty years, only about 20 stellar systems containing at least one Earth-like planet have been found. This rarity makes the search process akin to finding a needle in a haystack. However, the AI model developed by Davoult holds promise for pointing telescopes toward those potential “needles” much more intelligently. According to the research team’s calculations, directing observation efforts according to the model’s predictions could significantly increase discovery efficiency with the same available resources.
Continuing the search: Focusing on younger stars for longer
Davoult explains that the search does not end here, but also touches upon the types of stars that might be more suitable for the emergence of intelligent life around them. Stars smaller than the Sun—such as K-type orange dwarfs and M-type red dwarfs—have a much longer lifespan than our star, giving life a longer chance to evolve around them if the right conditions exist. Furthermore, discovering an Earth-like planet around these stars is relatively easier compared to sun-like stars because the habitable zone is closer to the star, and thus the planet’s orbital period is shorter, making the signals indicating its presence clearer in observation data.
Davoult’s team points out in their study that they created three groups of synthetic models representing hypothetical planetary systems that differ only in the mass of the central star. This single difference directly affects the mass of the protoplanetary disk around the star and consequently the amount of material available to form planets. The result was that each group showed differences in the number and characteristics of the formed planets, highlighting the importance of studying diverse types of stars in planetary research.
A powerful tool despite the challenges
The researchers acknowledge that comparing computer-generated synthetic models with real planetary systems is complex, and that their model is not yet perfect. Davoult says: »It is really hard to compare the numbers and patterns of planets predicted by our models with reality; we know our model is not complete«. But she adds: »However, if we look at the big picture at the level of the planetary system as a whole, I am convinced that it is a very powerful tool«. Meaning that the model may not be able to predict every fine detail, but it is capable of identifying general trends useful for guiding search operations.
The first steps in the astrobiology journey
Davoult emphasizes that finding these planets is merely a first step on the path to understanding the possibility of life beyond Earth. She says: »If we observe a planet within a certain stellar system, that doesn’t mean we have discovered all planets in that system«. She points out that there may be an Earth-like planet in the same system that is a bit farther from the star or too small for us to have detected yet. This is where the model she developed comes in: it takes known information about a planetary system and tells us whether there is a possibility of an undiscovered Earth-like planet in that system.
Davoult is currently focused on searching for rocky planets located within the habitable zones of their parent stars. She explains: »The first and most important step is simply discovering these planets and preparing a comprehensive database for them—even if we don’t yet have any idea about the composition of their atmosphere or whether they actually support life«. For scientists, merely finding these Earth-like worlds marks the beginning of a long journey to explore their characteristics and perhaps, in the future, signs of life on them.
Frequently Asked Questions
1. How many Earth-like planets have been discovered so far?
Despite the discovery of about 6,000 exoplanets, only about 20 stellar systems containing at least one Earth-like planet have been found, highlighting the rarity of these planets.
2. Why is searching around stars smaller than the Sun preferred?
Smaller stars like orange and red dwarfs live much longer than the Sun, giving life more time to evolve. Also, discovering planets around them is technically easier.
3. What is the “habitable zone”?
It is the range of distances around a star where liquid water can exist on a planet’s surface—not too close for water to evaporate, and not too far for it to freeze.
