How does artificial intelligence get "rewards"?

Question

Machine learning

How is the artificial intelligence algorithm built in which it receives "rewards"? What helps a neuron to understand what is right and what is wrong?

Answer 1

In the general case, any NN has an error function.
An error is the observed response of the NA and the desired response in a given situation.
That. globally, NN learning is minimizing the error function on the NN state space.
This is the real reward for the National Assembly.