Space is full of surprises, and AI is helping us find them faster. In April 2026, a team led by Princeton University reported that a machine-learning-assisted search of NASA TESS data found 11,554 planet candidates. Amazingly, 10,091 of them were new. The team studied 83,717,159 star light curves from TESS’s first-year full-frame images. A light curve is a record of how a star’s brightness changes over time. (arxiv.org)
TESS, short for the Transiting Exoplanet Survey Satellite, launched on April 18, 2018. It looks for tiny drops in starlight. These drops can happen when a planet passes in front of its star. This is called the transit method. Because the new search looked at many faint stars, it went beyond the usual official searches and more than doubled the number of known TESS planet candidates. (nasa.gov)
But there is an important point: a planet candidate is not a confirmed planet yet. It is a strong clue, not final proof. Scientists still need more observations with other telescopes to check whether each signal is truly a planet. The good news is that the new method already showed real success. Follow-up measurements confirmed one new world, a hot Jupiter orbiting the star TIC 183374187. (science.nasa.gov)
Why does this matter? NASA says scientists have found more than 6,000 confirmed exoplanets so far, and thousands more are still waiting to be tested. AI can help researchers study huge amounts of public NASA data much faster than people can by hand. So this story is not only about 10,000 possible new worlds. It is also about a new way of doing science. With smarter tools, we may discover many more hidden planets in the years ahead. Maybe one of these distant candidates will one day become a famous new planet, or even a world that could support life. (jpl.nasa.gov)










