Scientists are no longer only asking AI to explain experiments. They're asking it to invent them.

Here's the full story, via TU Wien.

As a student in Vienna, Mario Krenn's research group couldn't find a working configuration for a quantum experiment. So he described the available lab components mathematically and had an algorithm search for combinations that would work. "Programming it only took a few hours. Then I went home and left the computer running," he says. "When I came into the office the next day, the program had produced a file containing a proposed solution... unlike all of us, the computer had found an experimental setup that satisfied the necessary criteria."

This matters for understanding what's actually happening: "This approach has little in common with the kind of AI familiar from large language models. Chatbots are trained on enormous amounts of data and then generate solutions that are statistically likely." Instead, "it is an enormous optimisation problem... There is an overwhelmingly large space of possible experiments that can be built from the available components. The computer has to search this space systematically in order to find the best possible solution."

"This approach has already been used to improve fusion reactors, develop new ideas for particle detectors and generate proposals for making gravitational-wave detector systems even more sensitive." Philipp Haslinger, head of TU Wien's Center for Electron Microscopy, adds: "Artificial intelligence can therefore identify microscope designs that a human would probably never have come up with, but which can produce significantly better images or offer entirely new measurement possibilities."

"Sometimes you look at these computer-generated experimental proposals and quickly understand the idea behind them... But sometimes it is also very difficult to understand. You can calculate that the new experimental setup works better, but you cannot really put into words why." The goal, Krenn says, is "something like a universal physics simulator" that could predict outcomes across a much wider range of experimental setups.

"That is precisely the challenge: defining as accurately as possible what you actually want, and which constraints have to be satisfied, for example, a maximum cost, or a maximum amount of energy the device can absorb without exploding." Asked whether this means handing off science's "eureka moments" to machines, Krenn pushed back: "No, absolutely not. Human work is simply shifting to a higher level... using these tools will still require scientific expertise, creativity and a good intuition for physics."

Source: TU Wien: https://www.tuwien.at/en/all-news/news/kuenstliche-intelligenz-schlaegt-neue-physik-experimente-vor

Want practical AI guidance for parents and educators every week? Subscribe: https://www.aibyage.com/?modal=signup&utm_source=beehiiv&utm_medium=newsletter