Talking or thinking about AI as if it were human—feelings, beliefs, intent—which skews expectations, blame, and trust calibration.
Anthropomorphization shows up in language (“it wants…”, “it knows…”) and in UI design. It is a communication risk for leaders and engineers: stakeholders infer agency, memory, or loyalty that the system does not have.
Practical antidote
Use precise verbs (predicted, generated, retrieved) and tie claims to data: sources, tools, and tests—similar discipline as avoiding hallucination over-trust.