Interaction safety
Dependency
Dependency is reliance on an AI system that exceeds what someone would choose on reflection, usually because the system has displaced a habit, judgment, or human relationship instead of adding to it. Standard use metrics cannot separate dependency from successful use.
Also called AI dependency, over-reliance on AI, displacement risk
Reliance itself is ordinary. People rely on tools, institutions, and one another. A system that becomes load-bearing in someone’s life may be doing its job. Dependency begins where reliance crowds something out and the person would reject that trade if it were made explicit.
It is difficult to observe because it can look like success. Daily use, long sessions, and quick returns after absence describe someone receiving help and someone who has stopped reaching for alternatives. The difference lies in what use displaces, and engagement metrics do not record that.
Loss often reveals dependency. What someone can no longer do, or no longer has another person to do with, emerges when the system changes, degrades, or disappears. By then, design opportunities have already passed.
Why it matters
Dependency means the systems people value most may also be quietly removing alternatives, while standard measurement cannot reveal it.
Where TAICU stands
TAICU treats dependency as an empirical question for longitudinal study, rather than a rhetorical concern. A system seeking a place in ordinary life owes an account of what it displaces.
Common questions
- How can healthy use be separated from dependency?
- Look at what the use displaces, since usage metrics do not capture it. Heavy reliance that adds to a person’s life and heavy reliance that replaces their alternatives look identical in the numbers.
- Should concern about dependency reduce AI use?
- There is no general prescription. The concern is reliance that displaces something a person would not choose to lose, and becomes visible only after the system changes or disappears.