New AI masters games, emulates human brain

A new AI approach, Axiom, has demonstrated the ability to master simple video games more efficiently than current deep learning methods. This could lead to more capable AI agents. Axiom, developed by Verse AI, models the world like a human brain, using prior knowledge and active inference to predict game outcomes. It outperforms traditional AI by requiring fewer examples and less computational power. This new approach, inspired by the free energy principle, may be crucial for building AI agents that can learn how to act in the real world, according to experts. It offers an alternative to the dominant deep learning methods.