U.S. scientists have succeeded in creating a living computer by culturing more than 80,000 mouse stem cells (opens in new tab) (via IT Home) (opens in new tab). It is hoped that someday robots will be created that use living muscle tissue to sense and process information about their environment.

Researchers at the University of Illinois have used tens of thousands of living mouse brain cells to create a computer that can recognize patterns of light and electricity. The team presented their findings in the form of a palm-sized computer at the American Physical Society.

The scientists grew a computer consisting of 80,000 reprogrammed mouse stem cells and placed them between optical fibers on a grid of electrodes. During the experiment, the cells were kept alive in an incubator, and the team trained the mouse brain computer by repeatedly flashing 10 different patterns of electrical pulses for an hour, and after letting it rest for 30 minutes, recorded and analyzed the signals sent by the neurons. [The F1 scores (a measure commonly used for neural networks) were, frankly, not very high. It was measured on a scale from 0 to 1, with 1 being perfect recognition of the pattern. And initially, it could not score higher than 0.6 because of the random spikes of electricity from the neurons.

But Andrew Doe and his team at the University of Illinois found that by mixing chemicals and electron impulses, they could apparently quiet such randomness. The result was a rather eye-popping score, with the highest score being 0.98. So congratulations, strange mouse brain robot! [So-called reservoir computing matches neurons made from living cells with conventional computer chips to read the data they produce. The idea is that by dividing the labor between flesh and silicon, the time and energy required to train a conventional neural network can be reduced.

Ultimately, the hope is to use reservoir computing to create more efficient devices, and as New Scientist reports, at this point the mouse-shaped brainbot is still behind conventional neural networks in terms of raw performance. It falls short of traditional neural networks in terms of raw performance, but the research team aims to create a larger living computer.

Somewhat worryingly, the team also hopes that this large mouse-brainbot will exhibit behaviors that were not input into or trained by the neural network. I've read enough science fiction to know how that would work.

This is not the first time mouse cells have been used to power biohybrid technology in the wild. In January, scientists revealed that they used mouse muscles and 3D printing to create a soft scaffold to create a tiny robot that navigates a small maze.

One of the more realistic applications, according to New Scientist, is a robot that can eventually sense its environment and process data simultaneously. In other words, we are one step closer to carnivorous robots. The researchers also state that living cells for computing could make reservoir computing devices (used for the best machine learning algorithms) more power efficient and serve as a backup in case a component fails.

And I am not at all horrified by this. Wouldn't you?