Chapter 1: Human-AI Interaction

Human-AI Interaction

System behavior: Probabilistic
Capability Visibility: Fuzzy and emergent
Error Profile: Routine and expected
System Dynamics: Adaptive and shifting
Primary Design Goal: Calibrated trust
Mental models: Uneven competence

Human Computer Interaction

System behavior: Deterministic
Capability Visibility: Fixed and learnable
Error Profile: Bugs to fix
System Dynamics: Static
Primary Design Goal: Usability and efficiency
Mental models: Fixed accordances

Eliza Effect

Tendency for people to attribute understanding, intelligence, empathy, and intention to an AI system beyond what its actual capabilities justify.

Google Effect

Expecting information to remain externally available shifts memory from remembering the information itself toward remembering where to retrieve it. (cognitive offloading; transactive memory)

Distinctiveness of HAI

Non-determinism

Opacity

Calibrated reliance

Co-adaption

Yang et al.

Capability uncertainity + Output complexity