Amateur astronomers use AI to hunt for new exoplanets with varying success

Citizen scientists are leveraging AI models like Claude to analyze telescope data and identify potential exoplanet candidates, with some findings now under scrutiny by professional astronomers. One amateur, Pavel Rabtsevich, used an AI to detect two possible planets orbiting a star 116 light years away. While results are promising, the reliability of AI-driven discoveries remains mixed.
In September 2026, amateur astronomer Pavel Rabtsevich used an AI model—specifically Claude—to analyze astronomical data from a star located 116 light years from Earth. After providing the AI with a high-level directive, the system identified two potential exoplanet candidates within approximately two weeks. This marks a notable moment in the growing trend of citizen scientists leveraging AI tools to contribute to professional fields like astrophysics. While the findings are currently under review by professional astronomers, the event underscores the increasing accessibility of advanced computational tools to non-experts. The use of AI in this context builds on earlier efforts where machine learning was employed by research institutions to sift through vast datasets from telescopes like Kepler and TESS, but now the frontier is expanding to include individuals without formal training. The broader context includes prior reports of AI uncovering hidden patterns in data, such as decoding historical ciphers, suggesting a pattern of democratized discovery powered by accessible models. The individuals involved—amateurs like Rabtsevich—seek recognition and scientific contribution, while organizations such as NASA and academic observatories may benefit from expanded data analysis capacity. However, the reliability of AI-generated findings remains uncertain, and the scientific community has yet to validate the two candidates. What is not yet known is whether these signals are genuine planetary transits or artifacts of data noise, and how such discoveries will be formally vetted in the future.
The involvement of amateur astronomers using AI to detect exoplanets could reshape how scientific discovery is distributed, potentially lowering barriers to entry in astronomy and fostering public engagement with science. If validated, such findings may encourage more individuals to participate in data-driven research, possibly leading to new collaborative models between citizen scientists and professional institutions. This trend may also influence funding and policy decisions, as agencies consider supporting platforms that integrate AI with public participation. A comparable earlier case is the success of the Galaxy Zoo project, where volunteers classified galaxies and contributed to major discoveries—though without AI assistance. The current scenario, however, introduces a new layer: AI as a co-investigator, raising questions about attribution, reproducibility, and the role of human oversight. Institutions may need to develop new protocols for verifying AI-generated hypotheses, particularly as models like Claude become more autonomous. The broader impact could extend to education and public trust in science, as people witness real discoveries emerging from non-traditional sources. What to watch for next is whether professional astronomers adopt standardized frameworks for evaluating AI-assisted findings, and whether such tools become embedded in citizen science platforms. This could either enhance scientific output or introduce new challenges in data integrity and peer review.