AI Discovers Hundreds of Cosmic Oddities in Hubble Space Telescope Data
The Hubble Space Telescope has captured snapshots of our universe for over 35 years. This extensive collection of images holds countless cosmic treasures waiting to be identified. Recently, a team of astronomers used an advanced artificial intelligence (AI) tool to scan nearly 100 million Hubble images. Their effort uncovered more than 1,300 unusual space objects, including some never seen before!
How AI Speeds Up Space Exploration
Tackling millions of astronomical images is a huge challenge. Until now, scientists examined many images by hand or with the help of citizen science projects. Although helpful, these methods were too slow to keep up with the vast Hubble data archive.
AnomalyMatch: The AI Eye in the Sky
The new AI tool, named AnomalyMatch, was created by David O’Ryan and Pablo Gómez at the European Space Agency (ESA). It uses neural networks that mimic how the human brain processes visual information. This capability allows AnomalyMatch to detect rare and strange objects across millions of images quickly.
A Massive Search for Strange Objects
The team taught AnomalyMatch what “normal” celestial objects look like by training it on well-known space data over the course of its training. After that, it examined millions of extremely small image patches that were taken from the Hubble Legacy Archive of NASA. The expedition uncovered approximately 1,300 peculiar and fascinating things in just two and a half days, of which more than 800 were brand new to the scientific community.
The Cosmic Oddities AI Found
The unique discoveries include different types of galaxies and other rare celestial phenomena:
Galactic Mergers and Interactions
Many detected galaxies are colliding or pulling on each other gravitationally. These interactions cause strange shapes and long trails made of stars and gas.
Gravitational Lenses: Nature’s Magnifying Glasses
Some galaxies act like giant lenses due to their gravity bending light from background galaxies into arcs or rings. These effects help astronomers study distant parts of the universe.
Unusual Galaxy Forms & Planet-Forming Disks
The artificial intelligence also discovered jellyfish-shaped galaxies with gaseous tentacles that extended outward. Furthermore, researchers observed edge-on planet-forming disks that resembled hamburgers in our Milky Way galaxy. Photography rarely captures these objects in a clear manner.
The Future Role of AI in Astronomy
The tremendous advancement that has been made shows how technologies that utilize artificial intelligence could potentially benefit scientific research. As a result of the introduction of this technology, researchers are now able to analyze historical data in a more expedient and comprehensive manner than ever before.
Bigger Data from Upcoming Telescopes
Forthcoming missions like NASA’s Nancy Grace Roman Space Telescope, the ESA’s Euclid, and ventures like the Rubin Observatory will soon generate a significant increase in space data.
Tools that are similar to AnomalyMatch will become indispensable for the rapid analysis of these enormous archives, which will also assist in the discovery of unexpected occurrences that are currently unknown to us.
Tackling Growing Data Volumes
Soon, telescope missions like NASA’s Nancy Grace Roman Space Telescope and the ESA’s Euclid will collect enormous volumes of data. This inundation renders human examination impossible without AI assistance.
A Hybrid Approach to Discoveries
In order to ensure accuracy, the researchers utilized both automatic searches and manual verification performed by experts. Using this innovative approach, the search is significantly sped up while yet keeping its scientific integrity.
Additionally, to stay updated with the latest developments in STEM research, visit ENTECH Online. Basically, this is our digital magazine for science, technology, engineering, and mathematics. Further, at ENTECH Online, you’ll find a wealth of information.
Reference
- O’Ryan, D., & Gómez, P. (2025). Identifying astrophysical anomalies in 99.6 million source cutouts from the Hubble Legacy Archive using AnomalyMatch. Astronomy and Astrophysics, 704, A227. https://doi.org/10.1051/0004-6361/202555512



