Everyday life, relationships and psychology

• Anthropomorphism: because it speaks so well, we attribute to AI an understanding, intentions, a consciousness that it does not have. This illusion can distort our trust and our decisions.
• Calibrating trust: distrusting too much deprives us of the benefits; trusting too much exposes us to errors and hallucinations (chapter 4). The aim is a well-calibrated trust, which requires understanding the tool's limits.
• Cognitive deskilling: systematically delegating writing, calculation, memory or reflection to AI can, over time, atrophy these faculties (the same risk as in education, chapter 15). Thinking with AI strengthens us; having it think in our place weakens us.

---

Key takeaways (chapter 19)

• AI is a daily and often invisible presence (recommendations, assistants) that shapes what we see and desire, at the risk of capturing our attention.
• AI companions offer real comfort but create parasocial relationships and a dependence; AI simulates the relationship, it does not experience it.
• In mental health, AI can help at the margins but carries serious dangers (inappropriate advice, poor handling of crises, documented harm); it does not replace a professional and must steer toward human help.
• Underlying effects for everyone: anthropomorphism, the need for well-calibrated trust, and the risk of deskilling. The principle: think with AI, not in its place.

From the intimate to the conflictual: chapter 20 examines AI as a weapon and as a shield, in the field of cybersecurity.

---