Neural networks, the architecture behind LLMs, took loose inspiration from how neurons connect and fire: input, processing, output.

You see a red light. You press the brake.

LLMs weren’t explicitly designed to imitate the brain, they emerged from a narrower goal: predicting the next word. But the parallel holds up anyway.

Email received → email response drafted.

Yet there are things the human brain does so elegantly that LLMs still don’t have.

1. Continuous, real-time learning — every experience updates the brain as it happens

2. Embodied emotion — feelings are generated through the body and shape what we notice, remember, and decide

3. Persistent memory — one continuous thread of experience carried across a lifetime