The human brain gives knowledge to Artificial Intelligence, enabling it to replicate some of the brain’s functions and learning processes. However, the extent of emulating the activity of the brain is not fully understood.
Researcher Hiroshi Makino recently did a study to identify the possible similarities between the brain and artificial intelligence in solving problems.
AI agents take pre-training abilities they have learned and recombine these skills in a hierarchical manner when tackling a new challenge. Makino was interested in investigating the notion that the brain functions similarly.
To accomplish this, he examined how deep reinforcement learning bots picked up new task-solving skills. Then he contrasted this with how mice approached the same novel assignment.
Similarities between Artificial Intelligence and The Brain
Makino investigated the neural activity of the mouse cortex, as the mice undertook a novel task using previously learned skills. Furthermore, as he approached a new composite problem. He noticed brain representations of action values that mirrored those produced by a deep reinforcement learning algorithm.
The findings of the study conclude that when a new task is being tackled, the brain consists of a new behaviour. And it is through a simple arithmetic operation on action-value representations with scholastic policies acquired previously.