# AI-Powered Discovery Uncovers 10,000 New Exoplanet Candidates

Princeton-led team breaks records in space exploration with machine learning breakthroughs.

- **Source:** [Forbes - Business](https://brifing.si/source/e0310925-a7af-4d4b-80a7-c80d400e49fa)
- **Published:** 2026-07-27
- **Topic:** [Business](https://brifing.si/c/business)
- **Original article:** <https://www.forbes.com/sites/kevinholdenplatt/2026/07/27/super-speed-ai-helps-astronomers-find-cosmic-trove-of-new-exoplanets/>
- **On Brifing:** https://brifing.si/s/0996f2dd-c417-4612-a11f-0bc7dcff15e3

## The Era of AI-Assisted Discovery

A team led by Joshua Roth at Princeton University utilized advanced artificial intelligence tools to analyze data from NASA's TESS space telescope. By scanning 80 million light curves, the team identified over 10,000 new planet candidates. This discovery marks the largest haul since the first detection of a planet orbiting a sun-like star in 1995. Specialized software like the Cambridge Exoplanet Transit Recovery Algorithm accelerated the data analysis process dramatically.

## Diversity of Planetary Systems

The candidates include over 9,000 'Hot Jupiters' (gas giants in close orbits), more than 100 Neptune-sized planets, and 11 potential 'Super-Earths'. Hot Jupiters are easier to detect due to their large size and short orbital periods. These planets orbit various star types from M-dwarfs to blue supergiants, showcasing our galaxy's diversity. Particularly intriguing are Super-Earths in habitable zones, which remain a primary focus for researchers.

## Global Collaboration

An international consortium including MIT, UCLA, and Chile's Las Campanas Observatory collaborated on analyzing TESS data. The Random Forest machine learning model reduced 80 million light curves to 2.5 million, cutting analysis time from 3.5 years to 6 weeks. This methodology has attracted attention from ESA and NASA, while inspiring similar initiatives at institutions like Cambridge University.

## Technological Breakthrough

Roth's team aims to implement convolutional neural networks (CNNs) by late 2026 to eliminate manual vetting entirely. The 10,000 samples from TESS's first cycle will train these AI models. With next-generation space telescopes coming online, data processing speeds are expected to increase further. These developments suggest an explosion in habitable planet discoveries within the next decade.

## Anatomy of Discovery

The TESS telescope records light dips caused by planets transiting their host stars. AI algorithms identify patterns in these transit signals to flag potential planets. While Hot Jupiters block 1-3% of starlight, Earth-like planets cause merely 0.01% dips. The team employs specially trained deep learning models to detect these subtle signals that would be impossible to spot manually.

## Milestone in Astronomy

This discovery demonstrates AI's transformative power in scientific research. The system completes in weeks what would take years through traditional methods, particularly revolutionizing 'rare event' detection. Roth calculates the AI reduced processing time by 32x. NASA's TESS telescope, launched in 2018, surveys 20x more stars than previous missions with its 400x wider field of view.

## Future Projections

The team is now analyzing TESS's second cycle data, hoping to find more Earth analogs by adjusting search parameters. Rocky planets in stellar habitable zones (Goldilocks zones) with liquid water potential are primary targets. Next-gen instruments like the James Webb Space Telescope will examine atmospheric compositions of these candidates for biosignatures.

## Scientific Legacy

Building on Princeton's legacy of 20+ Physics Nobel laureates, the team pushes boundaries in AI-assisted astronomy. Roth's doctoral research pioneers methodologies applicable not just to planet hunting but also dark matter studies and galaxy formation research. These findings represent a critical step in humanity's quest to understand our place in the cosmos.

## Sources covering this story

- https://www.forbes.com/sites/kevinholdenplatt/2026/07/27/super-speed-ai-helps-astronomers-find-cosmic-trove-of-new-exoplanets/ — 

---

_Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._