AI Discovers Rare Cosmic Objects in Hubble Images

For more than 35 years, the Hubble Space Telescope has amassed a staggering archive of astronomical images and data, capturing the universe in unprecedented detail. Analysing this wealth of information manually, however, is painstakingly slow. Each image requires careful scrutiny to uncover subtle cosmic phenomena, and with billions of exposures in its archive, human researchers simply cannot keep pace.

Recently, a team of scientists at the European Space Agency (ESA) employed an artificial intelligence (AI) system to tackle this challenge. The AI, known as AnomaliMach, scanned nearly 100 million Hubble images in just two and a half days, identifying over 1,300 unusual and rare cosmic objects, many of which had never before been catalogued.

AnomaliMach: AI for the Cosmos

Developed by ESA researchers David O’Ryan and Pablo Gomez, AnomaliMach is a neural network designed to emulate human cognitive processes. The system was trained specifically to recognise rare cosmic structures, such as jellyfish galaxies and gravitational arcs, which are otherwise difficult to spot amid vast data sets. After its training, AnomaliMach scanned Hubble’s legacy archive in just 60 hours, generating a preliminary catalogue of anomalous objects for human verification.

Upon review, astronomers confirmed that the AI had indeed detected 1,300 genuine anomalies, of which over 800 had not been previously documented in scientific literature.

Key Discoveries

Among the notable discoveries were galaxies caught in the act of collision. These interactions produce elongated stellar tails and dramatically distorted shapes. The AI also identified several jellyfish-like galaxies, whose trailing gas streams give them their name. In addition, a few dozen objects could not be classified into any known cosmic category, highlighting the potential for entirely new astrophysical phenomena.

Accelerating Future Space Research

According to the researchers, upcoming missions such as ESA’s Euclid and NASA’s Nancy Grace Roman Space Telescope will generate unprecedented volumes of data. In this context, AI will be indispensable for rapid analysis.

As Gomez emphasised, “AI is not replacing astronomers; it acts as a tireless assistant. It quickly scans data, flags objects of interest, and lets human experts make the final judgements.” The synergy of computational speed and human insight promises to unlock the universe’s deepest mysteries more efficiently than ever before.

In total, this study demonstrates that combining Hubble’s rich archives with cutting-edge AI offers a transformative approach to cosmic discovery, potentially revealing phenomena that have remained hidden for decades.