A series of promising designs for new gravitational wave detectors have been developed with a helping hand from artificial intelligence, which have left scientists with new questions about the cosmos.
Urania, an AI developed by a team led by Dr. Mario Krenn of the Artificial Scientist Lab at the Max Planck Institute for the Science of Light, produced the new detector designs, which they say have already shown promise to exceed the capabilities of existing human-made concepts.
Intriguingly, Urania did more than merely replicate existing strategies scientists rely on for the detection of these cosmic features first predicted by Einstein more than a century ago. According to a statement released by Krenn’s lab, the AI created entirely new ones, “some of which scientists don’t yet fully understand.”
The team’s results could mark the beginning of a future where intelligent machines commonly produce new discoveries about our cosmos, leaving the remaining job of interpreting them to humans.
From Einstein’s Vision to AI’s Imagination
Physicist Albert Einstein initially predicted the existence of gravitational waves in 1916. One century later, scientists finally succeeded in directly detecting them with the aid of detectors developed by the Laser Interferometer Gravitational-Wave Observatory (LIGO) project.
Working alongside LIGO researchers, Krenn and his team hoped to expand on current concepts for gravitational wave detector designs by employing artificial intelligence.
Specifically, the team aimed to produce cutting-edge new gravitational wave detector designs involving interferometry, which measures very subtle changes in how waves interfere with each other once they interact.
The result of this collaboration was Urania, which can produce such next-level interferometric detector designs—a demanding process, which involves many layers of work that include dialing in the parameters of the detectors, as well as optimizing their layout.
By using machine learning to help remedy the need for ongoing optimization, Urania was able to assist the team in creating a range of novel experimental designs for gravitational wave detectors.
New “Alien” Detector Designs
Remarkably, some of the AI’s new designs appear to outperform even some of the best existing concepts for detectors that physicists hope to build in the coming years, greatly improving their sensitivity and widening the overall detectability range of the target gravitational wave signals.
While much of Urania’s concepts revealed known processes upon analysis, some of the AI’s designs were very unconventional, and suggested new applications that scientists don’t yet fully understand, and which may potentially help shape the next generation of detector technology.
“After roughly two years of developing and running our AI algorithms, we discovered dozens of new solutions that seem to be better than experimental blueprints by human scientists,” Krenn said in a statement.
Krenn and the team say that they have assembled a collection of 50 of the best performing designs and made them publicly available in a “Detector Zoo” to promote further research, although many of the “tricks” the AI appears to have used in creating these novel designs remain a bit baffling, for now.
“We are in an era where machines can discover new super-human solutions in science, and the task of humans is to understand what the machine has done,” Krenn said of the work his team and Urania have accomplished.
“This will certainly become a very prominent part of the future of science,” Krenn added.
The team published their findings in a paper, “Digital Discovery of Interferometric Gravitational Wave Detectors,” published in Physical Review X on April 11, 2025.
Micah Hanks is the Editor-in-Chief and Co-Founder of The Debrief. He can be reached by email at micah@thedebrief.org. Follow his work at micahhanks.com and on X: @MicahHanks.