Superintelligence

Why it’s on the list: The book that put the alignment problem in front of a general audience: Bostrom works through how a machine intelligence might come to exceed human capability, why an arbitrary objective plus great capability is dangerous by default, and what a controlled transition would have to look like.

Where I saw it: A bare Amazon link saved in the note.

Connections

• Nick Bostrom — author.
• AI Safety · Existential Risk — the fields this book largely defined for a lay readership.
• Artificial Intelligence · Singularity — the subject.