Paper 7
How AI Works
Eleven concepts, from neurons to Da Vinci
- Authors
- Zachariah Roydel Robertson
- Date
- 2026-10-08
- License
- License not stated
Abstract
Eleven concepts that power modern language models, in the order you need them: from a single artificial neuron — weights, backpropagation, tokens, next-word prediction, transformers and attention, pretraining, fine-tuning, reinforcement learning, test-time thinking, open and closed weights — to how Da Vinci learns: stochastic training, causal reasoning and self-improvement on open-weight foundations.
Keywords
- Neural networks
- Transformers
- Pretraining
- Reinforcement learning
- Da Vinci