Cybernetics provides both positive and negative feedback for ai agents to become stable

08.14.2019 - By Machine learning

Download our free app to listen on your phone

In Cybernetics, Wiener established the universal principle of feedback. Negative feedback, as described by Wiener was not such a negative thing at all. Negative feedback was error correcting information yielding order and self-control to automatic machines, to the body and the brain, and to people in their daily lives. Positive feedback was information that confirms and reinforces the outcome of communication or control process. However, when positive feedback is left uncheck it can lead too runaway processes, wild oscillations, and ultimately something catastrophic. Wieners favorite example, “in the plight of hapless drivers on icy road who can veer out of control with one tap on the accelerator, and whose brakes are useless to provide the crucial negative feedback required to stop the car.”

More episodes from Machine learning