AI is Replacing Physicists

The experiment, featuring the small red glow of a Bose-Einstein condensate trapped in infrared laser beams. (Image courtesy of Stuart Hay/ANU.)
How many people and years of training does it take to replicate a Nobel-Prize-winning physics experiment? Apparently none.

Researchers recently used an artificial intelligence to run a complex experiment, which it learnt to perform from scratch in under an hour. "A simple computer program would have taken longer than the age of the Universe to run through all the combinations and work this out," said co-lead researcher Paul Wigley from the Australian National University Research School of Physics and Engineering.

This suggests that even physicists are on track to having their jobs augmented if not outright captured by artificial intelligence.


The Coldest Place in the Universe

The experiment involved the creation of a Bose-Einstein condensate, an extremely cold gas trapped in a laser beam. At a billionth of a degree Kelvin, it is even colder than outer space. To maintain this temperature, the AI took control of the three laser beams used to cool the gas.

The researchers were surprised by the novel methods it employed.

Researchers Paul Wigley and Michael Hush with their experimental apparatus. (Image courtesy of Stuart Hay/ANU.)
For example, it fine-tuned the power settings of the laser beams to reduce overall power expenditure in a way no human could. It was also able to quickly recalibrate and compensate for any fluctuations between experimental trials. Again, not something a human can do.

Since Bose-Einstein condensates are extremely sensitive to external disturbances, this AI system could be suitable for field measurements that require extreme precision, such as those in mineral exploration and navigation. It could even be used to measure minute changes in the Earth’s gravity or magnetic field.

"You could make a working device to measure gravity that you could take in the back of a car, and the artificial intelligence would recalibrate and fix itself no matter what," said Dr. Michael Hush from the University of New South Wales.

What's more, it's portable and cheaper than a physicist. 

The research is published under the title, “Fast machine-learning online optimization of ultra-cold-atom experiments” in Scientific Reports.