Non-invasive command neuro interfaces are losing their practical relevance for generations and must be replaced by systems based on the interaction of the brain and artificial intelligence. A neurophysiologist, one of the founders of neurointerface science in Russia and a professor at Lomonosov Moscow State University, told socialbites.ca. Lomonosov Alexander Kaplan.
“Generation 5.0 neuro interfaces should work not to block brain commands like in generations 1.0-4.0 but to create a channel of interaction between the brain and AI elements, such as various artificial neuron networks. “This is an attempt to get a channel to teach artificial intelligence about the characteristics of the human brain,” he said.
Kaplan explained that modern combinations of neural networks can learn literally anything from scratch. However, it is important to create the conditions under which the neural network will understand exactly what it needs to learn.
“In neural interface 5.0, there will be two combinations of neural networks: circuit 1 and circuit 2. Circuit 1’s job will be to learn from the encephalogram to understand when the brain is accepting in response to the suggestion made and when it is not. Even better, teach the grid to understand this in four stages: “yes”, “almost yes”, “almost no” and “no.” In the second circuit, the neural networks not only analyze the current EEG, they make assumptions based on it. If a person is up to something, then what exactly?
For more on why previous developments in neural interfaces have not justified themselves, see material “socialbites.ca”.