Amateur astronomer uses AI tools to find possible exoplanet and wins TESS observing time

A 28-year-old product manager in Spain used multiple AI coding tools to analyze public data from NASA's TESS space telescope and identified periodic dimming in a star called TIC 4206066 that is consistent with a possible exoplanet, according to Gizmodo. His request for TESS to point at the star was approved for an observation period running from October 31 to November 26.

The AI-assisted analysis found brightness dips every 3.18 Earth days in the star's data from 2018, 2020 and recent observations. Pavel Rabtsevich then ran 74 tests with Claude Code attempting to disprove the hypothesis, including one where the AI correctly predicted dimming times in a third dataset after being given two others.

Rabtsevich submitted a report requesting the TESS observation of TIC 4206066, and the request was approved. He has publicly posted the predicted dimming times for the observation window and said he would also post if the star does not dim on schedule.

Background

In the past, Rabtsevich had been combing through public TESS data looking for exoplanets. TESS, a space telescope launched by NASA in 2018, searches for exoplanets by measuring changes in stellar luminosity. The data he analyzed was taken from a narrow sliver of TESS's full range, captured from November of last year to early January. Astronomers have already cataloged more than 6,400 exoplanets, and TESS has identified more than 1,000 exoplanet candidates subsequently confirmed.