Home AI Scientists create a liquid computer that uses tiny moving particles instead of...

Scientists create a liquid computer that uses tiny moving particles instead of chips

Credit: DALLE. For illustrative purposes only.

What if a computer did not rely on electronic chips or tiny transistors, but instead used hundreds of microscopic particles moving through a liquid?

That futuristic idea has now become a reality in a laboratory, where researchers have created a completely different kind of computer that performs calculations using the natural movement of tiny particles.

The research, published in Communications AI & Computing, was carried out by scientists from the University of Konstanz and the University of Stuttgart in Germany.

Their work could one day help create computers that use much less energy while solving difficult problems.

Today’s computers process information using billions of tiny electronic switches called transistors. These switches rapidly turn on and off to perform calculations.

Artificial intelligence systems, including large language models, also rely on huge networks of simple computing units that require enormous amounts of hardware and electricity as they become more powerful.

The new system takes a completely different approach.

Instead of using electronic circuits, the researchers filled a liquid with several hundred microscopic particles. These particles constantly move in an oscillating, or repeating, pattern. When scientists feed information into the system, the particles interact with one another through the movement of the surrounding liquid.

As the particles move, they create tiny currents in the liquid that influence the motion of nearby particles. Together, they form complex patterns that naturally process the incoming information. Rather than programming every step of a calculation, the researchers allow the physical behavior of the particles to do much of the work.

This approach is based on a method called reservoir computing. In reservoir computing, a complex physical system transforms incoming information into rich patterns. Scientists then measure selected features of those patterns to produce the desired result. Unlike conventional computers, they do not need to understand every detail of how the system works internally, as long as it responds consistently.

The research team says they are the first to demonstrate reservoir computing using a large collection of interacting microscopic particles suspended in liquid.

To test the new computer, the researchers gave it difficult prediction tasks. One challenge involved forecasting chaotic time series. Chaotic systems, such as weather patterns, are extremely difficult to predict because even tiny changes can lead to very different outcomes. Despite this challenge, the particle-based computer made highly accurate predictions.

The system also proved capable of detecting extremely small changes hidden within noisy data. This is important because many real-world measurements contain a great deal of background noise that makes important signals difficult to find.

For example, scientists studying earthquakes often analyze seismic recordings that contain large amounts of unrelated vibrations. Climate researchers also examine complex datasets in search of subtle changes that could signal future environmental problems. A computer that can recognize these faint warning signs could become a valuable scientific tool.

Although the new computer is currently only a laboratory demonstration, the researchers believe it shows that useful computing does not have to rely on traditional electronics. Instead, complex physical systems can perform information processing on their own.

In the future, this idea could lead to a new generation of low-power computers and smart sensors that analyze information where it is collected, rather than sending everything to large data centers.

While practical applications are still some way off, the study opens an exciting new direction for computing by showing that even tiny particles moving through a liquid can work together to solve surprisingly complex problems.