Astronomers have taken a big leap forward in understanding black holes—thanks to the help of artificial intelligence.
A team led by Michael Janssen from Radboud University in the Netherlands has trained a powerful neural network to analyze millions of simulated black hole data sets.
Using this advanced tool, they’ve uncovered new insights into the supermassive black hole at the center of our Milky Way, known as Sagittarius A*, and its distant cousin in the galaxy M87.
In 2019, the Event Horizon Telescope (EHT) gave the world its first image of a black hole, located in galaxy M87.
In 2022, they released an image of Sagittarius A*. But those groundbreaking images were just the beginning. Beneath the surface of the data was a treasure trove of information too complex for traditional methods to fully decode.
That’s where the new AI tool came in. The team fed millions of synthetic black hole images into a Bayesian neural network—a type of AI that can not only make predictions but also estimate how confident it is in those predictions.
This is a major upgrade from earlier studies that relied on only a few sample models. By comparing real EHT data with a much wider range of simulations, the AI was able to pull out hidden patterns and make much stronger predictions.
One of the most exciting findings is that Sagittarius A* appears to be spinning extremely fast—almost at the speed limit for a black hole. Even more interesting, its spin axis seems to be pointing directly at Earth.
The team also found that the bright emissions near the black hole are likely caused by scorching-hot electrons in a surrounding disk of gas and dust, rather than by a jet of material shooting out, which is often seen in other black holes. They also observed unusual magnetic behavior in the disk, suggesting that our current theories about how these disks work may need some updates.
Lead researcher Janssen said the use of AI is just the beginning. As more telescopes come online—like the Africa Millimeter Telescope currently under construction—scientists will have even more detailed data to test Einstein’s general theory of relativity under extreme conditions.
The research didn’t stop at our galaxy. The team also reanalyzed data from M87*, the black hole that starred in the first-ever photo. They found that M87* is also spinning fast, but not as fast as Sagittarius A*, and interestingly, it’s spinning in the opposite direction of the gas falling into it. This reverse spin may be a sign that M87* once merged with another galaxy.
This breakthrough wasn’t just about clever algorithms—it required a massive computing effort. The team relied on powerful systems across several countries, using tools like TensorFlow and high-performance computing facilities to process and manage the huge volumes of data.
These findings offer a clearer, more dynamic view of some of the universe’s most mysterious objects—and hint that we’re only beginning to uncover what black holes are really made of.
