Space Hobbyist Uses AI to Discover Hidden Planet in NASA Data

Pavel Rabtsevich never expected his hobby would lead to discovering a new world.

The Spain-based product manager spent his free time analyzing data from NASA's planet-hunting telescope, TESS. When his custom AI algorithm flagged something unusual about a star 116 light-years away, Rabtsevich decided to investigate.

The star, named TIC 4206066, was dimming for about two hours every three days. Rabtsevich dug through years of telescope data and found 22 more instances of the same pattern. If confirmed, the signal points to a rocky super-Earth about 1.4 times larger than our planet, orbiting so close to its star that surface temperatures reach 1,000 degrees Fahrenheit.

What makes this discovery remarkable isn't just the planet itself. It's how Rabtsevich found it.

He used multiple AI agents that wrote code, ran calculations, and checked each other's work. But Rabtsevich was clear about his role: "The research itself was mine: the question, the plan, the rules, the computing, the decisions."

The potential planet had actually appeared in NASA's systems back in 2020 but was rejected due to lack of clear evidence. Years of additional telescope observations made the signal much stronger. Pablo Mercador Perez, a member of the TESS team at MIT, confirmed the signal only became statistically detectable in January 2025.

Greg Laughlin, an astronomer at Yale who now works in AI, calls Rabtsevich's work "a great example of human-AI teamwork." Large language models like Claude and Codex have proven surprisingly effective at finding overlooked discoveries in astronomical data.

The Ripple Effect

Rabtsevich's discovery has inspired other amateur astronomers to dive into NASA's public datasets with AI assistance. Several are already reporting promising early results in their own searches.

Professional astronomers are taking notice too. NASA has added the star to TESS's high-priority observation list for November, when the spacecraft's orbit brings it back into view. Multiple professional teams have recruited Rabtsevich to share his methods and critique his findings.

The discovery demonstrates something profound about modern science. Powerful tools once available only to researchers at universities and government agencies now sit within reach of anyone with curiosity and determination. Public data archives from NASA missions contain countless undiscovered secrets waiting for fresh eyes.

Whether Rabtsevich gets official credit for the discovery or not, he's already won something bigger: proof that citizen scientists equipped with AI can contribute meaningfully to humanity's understanding of the cosmos.

The super-Earth orbiting TIC 4206066 will likely pass further tests and join the official census of known planets, expanding our map of worlds beyond our solar system one amateur discovery at a time.