Pre-training is a crappy evolution: the practically buildable substitute for the process that gave animals their built-in hardware.

scaling laws Whether more compute keeps buying more capability, and where the curve now bends -- pre-training, post-training, test-time.

reinforcement learning Training by reward: environments, verifiable tasks, value functions, and whether it works or merely beats what came before.

neural networks The architecture almost all modern machine learning is built on, and what changes when it gets deeper.