Foundations
Before any roadmap there are four questions worth not skipping, because the answers you give without thinking them through will govern everything that follows: what the phrase “artificial intelligence” names, where the idea that a machine might do something like thinking comes from, why it worked now and not before, and what kind of knowledge a system that learns from examples produces.
This subsection opens the chapter for a practical reason. The resource map teaches you to build models and the critical counterpart asks at whose expense they get built; neither one discusses what it is that is being built. That is the discussion here, and it is older than the field: Turing opened it in 1950 and it has not closed.
A record of the evolution. This material comes from chapter 1 of v1, rewritten and brought up to date. Three entries from that chapter were left out: A few little problems around AI, which now lives expanded in Thinking AI from the Global South; Reading list, replaced by Getting started and To learn even more; and the humour and complex-systems sections, which did not survive a rereading. Each entry here links at the end to its original version. v1 exists in Spanish only.