Neural networks, a form of artificial intelligence (AI), can learn to recognize patterns in data, which makes them useful for tasks like image recognition. They can analyze vast amounts of imaging data in a fraction of the time that a human would take. Thanks to its longevity, the Hubble Space Telescope has accumulated a 35-year archive ripe for harvesting.

A team of astronomers developed an AI tool named AnomalyMatch to comb Hubble’s archives for rare and unusual objects. The results: more than 1,300 objects with an odd appearance, hundreds of which had never been seen before.