The human brain is amazing.
Although he consumes the equivalent of just two bananas a day, that doesn’t stop him from performing incredibly complex tasks with impressive efficiency. But a team of researchers have devised a way to build a prototype artificial neuron made up of incredibly thin graphene slits housing a single layer of water molecules, according to a new study. published in the journal Science.
And, instead of electrons, this artificial neuron uses ions.
Just like your brain.
New artificial neuron uses ions, like the human brain
The ultra-high efficiency of the brain depends on a basic unit that we know and love as the neuron, which consists of a neuron with nanoscale pores called ion channels. These channels alternately close and open depending on stimuli, but the ion flows resulting from this process generate an electric current, which emits action potentials, which are the crucial signals that allow neurons to communicate with each other. . Artificial intelligence (AI) can do it too. But AI takes a lot more energy – tens of thousands of times more, to be precise. That is why the modern challenge for researchers in the field is to design and build electronic systems that rival the energy efficiency of the human brain.
The study of nanofluidics is relevant because it studies the behavior of fluids in channels less than 100 nanometers wide. The new study showed how an electric field could assemble the single layer of water molecules into elongated clusters, which develop a key property called the memristor effect: when the clusters retain some of the stimuli they recently received. Much like the human brain, the researcher’s design saw slits of graphene mimic ion channels, in addition to ion flows and clusters. Moreover, using numerical and theoretical tools, scientists have discovered a way to assemble these special clusters so that they reproduce the physical mechanism of emission of action potentials.
Artificial neurons could improve brain-computer interface research, for better or for worse
In other words, the ability to transmit information from one artificial neuron to another. This historic work is underway within a French team, working with scientists from the University of Manchester, UK. The next step is to experimentally prove that these new systems can perform basic learning algorithms which, in turn, can become a basis for providing electrical memory. recall via artificial neurons. Of course, the further replication of human brain activity also has potential importance in increasing neuronal functions. For example, Elon Musk’s Neuralink, which ultimately aims to implant computer chips in the brain and usher in a new era of “superhuman cognition,” when the intangible computational power of machine learning analysis is combined with (again) the unprecedented creative intuition of the human mind.
However, there are downsides, not all of them just potential. “Without proper regulation [of brain chips], your innermost thoughts and biometrics could be sold to the highest bidder, ”wrote philosopher and cognitive psychologist Susan Schneider in a Observer report. “People may feel pressured to use brainchips to stay employed in a future where AI outshines us in the workplace.” And, if brain chips like Musk’s Neuralink activate or enhance familiar smartphone activities, that data can be analyzed for signs of possible criminal digital activity with tools that are easily tricked into making a false positive and reporting you. highly stigmatizing crimes. It might sound like a stretch, but a recent announcement from Apple reveals that the line between privacy and public safety is rapidly shrinking.