For decades, astronomers have scoured the skies for distant planets through telescopes, hunting for a single needle in a vast cosmic haystack — but that search has just taken a dramatic turn. NASA’s powerful new AI model, ExoMiner++, has pushed past human limits to flag thousands of potential new worlds buried deep within massive space data archives. The staggering find suggests the universe may be far more crowded with planets than scientists once believed, and artificial intelligence is quickly becoming the ultimate tool for uncovering them. Keep reading to see how AI is transforming our search for the unknown.
Meet NASA’s ExoMiner++
Astronomers have already found more than 6,000 planets beyond our solar system, called exoplanets, orbiting stars far from the Sun. Many of these discoveries came from NASA’s space missions, including the now-retired Kepler telescope and its successor, TESS, which continues to scan the sky for new worlds. But even after years of searching, these missions have collected so much information that countless planets are still hiding in the data.
All of this space data is freely available in NASA’s archives, allowing scientists around the world to hunt for new planets in different ways. To speed things up, researchers at NASA’s Ames Research Center turned to artificial intelligence. In 2021, they introduced ExoMiner, an AI system that helped confirm 370 new planets by re-examining Kepler’s data.
Now, that technology has leveled up. The team has developed ExoMiner++, a more advanced AI trained using data from both Kepler and TESS. In its first major scan of TESS data, the upgraded system flagged about 7,000 possible new planets. These “planet candidates” aren’t confirmed just yet, but they’re strong leads that scientists will now study more closely using other telescopes, bringing us one step closer to discovering new worlds in our galaxy.
How ExoMiner++ works
ExoMiner++ is free for anyone to use and can be downloaded from GitHub, giving scientists around the world access to a powerful new tool for searching through TESS’s ever-growing library of space data. By making the software open to everyone, NASA allows researchers to double-check results, test new ideas, and uncover planets that might otherwise be overlooked.
“Open-source software like ExoMiner accelerates scientific discovery,” said Kevin Murphy, NASA’s chief science data officer at NASA Headquarters in Washington. “When researchers freely share the tools they’ve developed, it lets others replicate the results and dig deeper into the data, which is why open data and code are important pillars of gold-standard science.”
At its core, ExoMiner++ acts like a highly trained digital filter. It reviews signals that may show a planet passing in front of a star and determines whether those signals are likely caused by an actual exoplanet or by something else, such as two stars orbiting each other. “When you have hundreds of thousands of signals, like in this case, it’s the ideal place to deploy these deep learning technologies,” said Miguel Martinho, a Kellogg Brown & Root (KBR) employee at NASA Ames who serves as the co-investigator for ExoMiner++.
Although Kepler and TESS scan the universe in different ways, the data they collect work well together. TESS surveys almost the entire sky, focusing on nearby stars, while Kepler closely studied a smaller region in greater detail. This compatibility allows ExoMiner++ to learn from both missions and deliver reliable results. “With not many resources, we can make a lot of returns,” said Hamed Valizadegan, the project lead for ExoMiner and a KBR employee at NASA Ames.
What’s next for ExoMiner++
Looking ahead, the team plans to make ExoMiner++ even more powerful. While it currently analyzes lists of possible planet signals, future versions are being designed to spot those signals directly from raw space data. And with more missions on the horizon, the tool’s potential will only grow. NASA’s upcoming Nancy Grace Roman Space Telescope is expected to capture tens of thousands of planet transits, and like TESS, its data will be publicly available. Advances made with ExoMiner could play a key role in uncovering new worlds in that data as well.
Source: Astrobiology