They designed the new chip, if it works, it will be a fundamental change for artificial intelligence

Although artificial intelligence makes everyday life and work easier in many areas of life, it also has serious drawbacks. Namely: its operation consumes a lot of energy, which is a very significant problem in the age of global warming and the energy crisis. Researchers at Loughborough University in England have now come up with a solution that can help reduce this energy hunger.

The Interesting Engineering according to his report, the chip can process time-varying data directly in hardware instead of relying on software running on computers. The researchers claim that the solution can consume up to 2,000 times less energy than traditional calculation methods, but the amount varies depending on the method of use.

According to Pavel Borisov, a physicist at the university, this means that engineers can rethink the construction of systems related to artificial intelligence. The study on this is Advanced Intelligent Systems was published in the scientific journal

Taking inspiration from the functioning of the human brain, the researchers created a niobium oxide-based memristor that was able to perform tasks related to image recognition and prediction, among other things.

https://pulzus.hvg.hu/egeszseg/20260403_mesterseges-intelligence-laborlelet-orvos-paciens-kapotlas

The development was tested with a model called Lorenz-63. The model was originally created for a simplified description of atmospheric convection, but it became famous in connection with the phenomenon called the butterfly effect, which clearly shows the extreme way in which even the slightest change in the initial conditions can influence the result.

During the tests, the model was able to successfully predict the short-term behavior of the chaotic Lorenz system using the data processed with the memristor, as well as to reconstruct the missing data. It also performed basic logic operations, showing that the same device can support many different tasks.

The researchers hope that their work can contribute to artificial intelligence consuming much less energy in the future.

If you want to know about similar things at other times, like it Facebook.com/hvg.tech” title=”https://www.Facebook.com/hvg.tech” target=”_blank” rel=”noopener noreferrer”>the Facebook page of the HVG Tech section.

(Our opening image is an illustration.)