Most AI systems of the past decade have been based on supervised learning, trained using human-annotated datasets. They have been hugely successful, but have notable flaws. This type of AI is of little help in understanding how the brain works because animals, including humans, do not learn from labeled datasets. Biological brains gain a deep understanding of the world by exploring the environment. Scientists are beginning to explore machine learning algorithms for self-supervised learning, such neural networks that show similarities to how the brain works . Of course the work of the brain is not limited to self-supervised learning, it is full of feedback connections that existing self-learning AI lacks. AI models still have a long way to go.
This article is reprinted from: https://www.solidot.org/story?sid=72471
This site is for inclusion only, and the copyright belongs to the original author.