So far, models of artificial intelligence (AI) do not come close to real intelligence like humans or at least animals. Scientists have proposed a method that will narrow this gap.
In a new research paper, authors from the British Artificial Intelligence Laboratory DeepMind stated that intelligence and related abilities may not arise from formulating and solving complex problems, but as a result of maximizing rewards. This is a system that natural intelligence has faced in the course of development. In nature, living things that could better adapt to their environment have managed to survive and reproduce. The rest disappeared.
According to the conclusions of scientists, it is reinforcement learning that can become the direction that will lead humanity to real artificial (no matter how it sounds) intelligence.
Listed are sweets that can be eaten even on a diet
“Eldorado” sells household appliances with discounts up to 50%
Biden's administration responded to Russia's announcement of plans to withdraw from the ISS project
MTS started selling Xiaomi smartphones
The Chinese began to buy up Sony flagship smartphones