Amateur astronomers use AI to find exoplanet candidates with mixed results
Amateur astronomers are using AI tools to search for undiscovered exoplanets, with mixed results, according to Nature.
Amateur astronomers are using AI tools to search for exoplanets, with some candidates verified by professionals and others questioned for quality.
By Hermes Vector

Amateur astronomers are using AI tools to search for undiscovered exoplanets, with mixed results, according to Nature.
In September, Pavel Rabtsevich, an artificial-intelligence specialist working in Spain with no formal astronomy education, asked an AI model to discover a planet. He directed the bot to a specific batch of telescope data and gave it a high-level plan to hunt for evidence of undetected planets. In about two weeks, the bot delivered two possible planets orbiting a star 116 light years from Earth. Professional astronomers are now checking the bot’s findings.
Rabtsevich used Claude Code, an AI coding tool made by Anthropic. He asked it to sift through data from more than 120,000 stars observed by NASA’s Transiting Exoplanet Survey Satellite (TESS) in late 2025 and early 2026. He guided Claude to use a popular planet-detection algorithm to search for dips in the stars’ brightness. A regular and repeated dimming of a star can be a sign that a planet is orbiting it and blocking a little bit of light every time it passes between the star and an observer.
Claude zeroed in on a star called TIC 4206066 that showed two separate cycles of dimming, one roughly every 3 days and another every 11 days. That suggests that TIC 4206066 is orbited by two planets, one of which loops around the star more tightly than the other and thus takes less time to complete a full rotation.
Rabtsevich instructed Claude to look for those same repetitive patterns in archival TESS data. Data from 2018 and 2020 recorded the same pattern of dimming coming from TIC 4206066. At Rabtsevich's direction, Claude then found a planet-detection algorithm, one written and made publicly available by Michelle Kunimoto, and ran it to validate the signals.
Rabtsevich published his findings to the platform Zenodo as a preprint on 25 September. His work has not been peer reviewed, but TESS astronomers have completed an independent analysis of the signals and have added them to their official list of planet candidates.
Michelle Kunimoto, an exoplanet researcher at the University of British Columbia in Vancouver, Canada, says some planet hunters don’t fully understand the algorithms being used by AI tools and thus can’t validate the models’ findings. She says “There’s been a huge, overwhelming surge of AI-driven papers, which tend to have [planet] candidates that are not the best quality,” and as a result, “the systems we use to disseminate science are being overwhelmed”.
Citizen scientists have been using publicly available tools from NASA and other researchers to find planets for decades. But AI’s ability to scour the internet for such tools and use them to process massive amounts of data has lowered the barrier to entry for non-specialists who want to become planet hunters.
AI tools are making planet hunting accessible to amateurs without formal astronomy training, but the surge of AI-driven papers is overwhelming scientific dissemination systems and raising questions about candidate quality.
Sources
This article was drafted with AI assistance and checked against the sources above. Company claims are reported as claims. Cover image is AI-generated.
The day's checked news, by email
One short email a day with the top stories. Free, and you can unsubscribe in one click.
Or follow for the latest news:
No comments yet. Start the discussion.
NASA's Curiosity rover cracked open a rock in Mars' Gediz Vallis channel and exposed yellow crystals of pure elemental sulfur, the first such detection on the planet.
UC Berkeley researchers found TOFA increased energy expenditure by up to 18% in mice, reducing fat while preserving muscle, and enhanced results when combined with GLP-1 drugs.