Imagine a universe teeming with planets beyond our solar system, waiting to be discovered. That's exactly what NASA's AI is doing, and it's about to get even more exciting. Scientists have already uncovered over 6,000 exoplanets—planets orbiting stars other than our Sun—thanks largely to NASA's Kepler and TESS missions. But here's the mind-blowing part: there are still countless worlds hidden within the vast data these missions have collected. And this is where AI steps in, revolutionizing the search.
In 2021, a team from NASA’s Ames Research Center unveiled ExoMiner, an open-source AI tool that validated 370 new exoplanets from Kepler data. Now, they’ve supercharged it with ExoMiner++, a new version trained on both Kepler and TESS datasets. But here's where it gets controversial: while some argue AI might overlook subtle planetary signals, others believe it’s the key to uncovering exoplanets at an unprecedented scale. What do you think?
ExoMiner++ has already identified 7,000 potential exoplanet candidates from TESS data in its initial run. These candidates are signals that likely indicate planets but need further verification from telescopes. The tool is freely available on GitHub, empowering researchers worldwide to join the hunt in TESS’s ever-growing public data archive.
Why does this matter? As Kevin Murphy, NASA’s chief science data officer, puts it, “Open-source software like ExoMiner accelerates scientific discovery.” When tools are shared openly, it allows scientists to replicate results, dive deeper into data, and uphold the gold standard of science. This collaborative approach is why the exoplanet field is advancing so rapidly.
ExoMiner++ works by analyzing transit observations—moments when a planet passes in front of its star, dimming its light. It distinguishes these events from other phenomena, like eclipsing binary stars. “With hundreds of thousands of signals, this is the perfect playground for deep learning,” explains Miguel Martinho, co-investigator for ExoMiner++. Despite Kepler and TESS’s different observation strategies—TESS surveys the entire sky while Kepler focused on a smaller patch—their datasets are compatible, allowing ExoMiner++ to train effectively on both.
And this is the part most people miss: the next version of ExoMiner++ will go even further. Currently, it flags planet candidates from a list of transit signals, but the team is working on enabling it to identify signals directly from raw data. This upgrade will make it even more powerful for future missions, like the Nancy Grace Roman Space Telescope, which will capture tens of thousands of exoplanet transits. Just like TESS, Roman’s data will be publicly available, aligning with NASA’s commitment to open science.
“Open science isn’t just about better science—it’s about better software,” says Jon Jenkins, an exoplanet scientist at NASA Ames. “It’s why we’re making such rapid progress.” NASA’s Office of the Chief Science Data Officer leads these efforts, ensuring scientific data, tools, and software are shared publicly to maximize the impact of their missions.
So, what’s next? With ExoMiner++ and future missions, the possibilities are endless. But here’s a thought-provoking question: As AI takes the lead in exoplanet discovery, how will it reshape our understanding of the universe? Will it uncover Earth-like worlds, or something entirely unexpected? Let us know your thoughts in the comments—we’d love to hear your take on this cosmic quest!
