A new machine learning approach tries to better emulate the human brain, in hopes of creating more capable agentic AI.

The idea is that graph networks are bigger than any one machine-learning approach. Graphs bring an ability to generalize about structure that the individual neural nets don't have.