AI reading noteStuart Russell & Peter Norvig
Artificial Intelligence: A Modern Approach
A broad map of AI through agents: how they search, reason under uncertainty, learn, perceive, and act. The useful shift is to define an agent by its goals and environment before choosing a technique.
Artificial Intelligence: A Modern Approach
Key ideas
- Formulate the task and performance measure before selecting an algorithm.
- Search, probability, learning, and decision making solve different parts of an agent problem.
- A capable system also needs to account for uncertainty, safety, and human values.
How I would apply it
Write the environment, observations, actions, and success measure for an AI feature before building it.