← AI intuition & LLM systems

Adjacent ideas worth recognizing

A broad map has room for ideas outside the main serving path. Go deeper when the workload makes them relevant.

Causality and uncertainty

Learn why predicting an outcome is different from predicting an intervention.

Search and control

Recognize symbolic planning, black-box optimization, and physical decision-making.

New sequence and generation families

Ask what information the model stores and how many steps it needs to produce a result.

Interpretability and scientific judgment

An explanation is a hypothesis to test, not automatic proof of a causal mechanism.