Augmented Mind: A Method for Learning, Working, and Building with AI - Volume 2: Trust, but Verify (Phương pháp học, làm và kiến tạo cùng AI)

Volume 2 of the Augmented Mind series turns the reflex of verifying before trusting into a teachable, practiced skill. Across seven chapters it explains why a language model fabricates even at its most confident, distinguishing fabricated facts, fabricated citations, and fabricated reasoning; how to spot hallucination through cross-checking, source-grounding, and consistency tests; how to calibrate trust per task while avoiding the symmetric traps of automation bias and excessive doubt; a repeatable verification process with provenance discipline; recognizing and unpacking bias in model output and rejecting the claim that AI is objective; and the signals for when to stop trusting. The volume does not tell readers to stop using AI but to keep final judgement in their own hands, warning against letting AI grade its own work. It builds on the structured dialogue of Volume 1 and feeds the evaluation discipline of Volume 3. No programming is required.