One of the distinguishing features of machines is that they don’t need to sleep, unlike humans and any other creature with a central nervous system.
However, one day, your toaster may need an occasional nap, just like your car, refrigerator, and just about anything else. Because, they were revolutionized with the advent of artificial intelligence (AI).
The change will come when AI systems that simulate living brains are incorporated into a range of devices that currently rely on computers and conventional microprocessors. This is the same research being conducted at the Los Alamos National Laboratory.
The network system may need a nap.
“We are investigating how these simulated networks respond to unsupervised training. In this type of activity, the network system sets out to classify groups without prior examples for comparison,” said Garrett Kenyon, the researcher who carried out the project.
According to this scientist, it will not be surprising if the network system becomes unstable after the time children use it for continuous learning. However, when exposing the networks to states similar to the waves experienced by the living brain during sleep, stability was restored. This is like the network system having a nap.
According to scientist Kenyon, this kind of instability is not characteristic of all AI networks. Most machine and AI researchers never encounter this instability. Because, in the artificial systems they study, these scientists can perform math that is not equivalent to living neurons.
Research results also show that, in both artificial and natural intelligence systems, slow-wave sleep can work. This ensures that the nerve cells remain stable and free from hallucinations.
“The dormant state in a neural network is very different from the one your computer has after a period of inactivity. A typical computer that effectively enters a “sleep” state will be in a suspended state. Then, all computation is frozen over time.
And popular advice is to try “turning off your computer and then back on” when they fail. This phenomenon is similar to causing your computer to brain dead for a short time,” explains researcher Kenyon.
In neural networks as well as in living organisms, sleep is not inactivity. It is another form of activity that is important for the normal functioning of nerve cells.
The researchers hope that the sophisticated AI system will help to better understand sleep and other features in biological systems.