The AI Interaction Risk Screen

The AI Interaction Risk Screen is a structured, practitioner-facing instrument for assessing how a person's interactions with an AI system are functioning in their life — built on the principle that risk lives in the person's trajectory, not in the technology category.

Its central axis distinguishes displacement from supplementation: whether the system is standing in where human support exists and is being crowded out, or supplying support in a domain where none was available. The two carry different risk profiles and warrant different responses, and the screen assesses them domain by domain rather than treating "AI use" as a single exposure. Version 1.1 — revised following structured peer review of v1.0 — introduced this axis alongside the Change-Over-Time Log, a longitudinal spine for repeated administration: the instrument's working assumption is that adaptive engagement and emerging dependency are distinguished over time, not from any single observation.

The screen deliberately stays at the functional, observable layer. It does not diagnose, and it does not measure mechanism-level cognition; it is designed to pair with clinical judgement and formal assessment, not replace them.

It is written from sixteen years of frontline mental health nursing practice, for clinicians, practitioners, researchers and services who need a repeatable way to watch the thing that matters: not whether someone uses AI, but what the use is doing to them over time. Part of the Heartbeat Framework series.