Create robots that “feed” on living neurons to think like humans

At this point, it is impossible to think that artificial superintelligence, as the successor to the internet boom, will not be the dominant paradigm of this century. A new study published in the journal Applied Physics Letters shows this. Three researchers from the Graduate School of Information Science and Technology at the University of Tokyo managed to “feed” a robot with a neuron culture produced from living human cells, which sent complex thought patterns through electrical stimuli.

The concept is based on the so-called neuromorphic computing. Its genesis lies in teaching machines with artificial intelligence to adopt human logic to solve all kinds of problems. In this case, it was observed whether a robot could escape a maze by memorizing the routes and recognizing the environments.

That neuromorphic computing began to see its first lights with the ideas of Carver Mead, an American electrical engineer at the California Institute of Technology (Caltech). He designed algorithms and circuits capable of mimicking the behavior of the animal nervous system in the 1960s.. But it wasn’t until the 1980s that the scientific community viewed this project with less skeptical eyes.

Despite these advances, reaching the complexity of the synapse as such will take longer. According to the neuropediatrician María José Mas, who works at Fontanet Medicine and Physiotherapy in Tarragona (Spain), and the Rambla Nova Medical Center, the synapse is the junction between a neuron and another cell.

There are two types: electrical, which propagate signals between neurons by charged molecules; and chemical, in which human cells do not establish direct contact, since the processes occur through a mediator called neurotransmitter.

The American Institute of Physics (AIP) released a press release stating that if the robot deviated in the wrong direction towards the end of the maze, the neurons were disturbed by electrical fluids. These interruptions did not stop until the machine was successful.

“I was inspired by our experiments myself to hypothesize that intelligence in a living system arises from a mechanism that extracts a coherent exit from a disorganized or chaotic state,” stated co-author Hirokazu Takahashi.

Takahashi also compared the learning of robots with that of an elementary school child who cannot yet solve mathematical problems posed to a university student, because “a richness of the repertoire of patterns” is lacking. So helping machines move up to the next evolutionary rung is related to giving them tasks as they pass various tests.

“The team believes that the use of physical deposit computing in this context will contribute to a better understanding of brain mechanisms and may lead to the novel development of a neuromorphic computer,” the researchers wrote in the AIP statement. The new era of robotics and its artificial intelligence is approaching by leaps and bounds.