Interaction safety
Anthropomorphism
Anthropomorphism is attributing human traits, including intention, feeling, understanding, and care, to a non-human system. Conversational AI reliably invites it through fluent language. People use these systems under that condition, and talk alone will not remove it.
Also called humanization of AI, attributing mind to machines
Telling someone that a system does not understand them changes little. The interaction produces the attribution because the system responds fluently, seems to track what was said, and adapts to the person. Human social cognition reads intention into that input as a normal response to something it never evolved to handle.
Disclaimers are weak tools here. A notice saying a system is AI gets read once, then thousands of exchanges can contradict its feel. Behavior speaks more clearly: what a system says about itself during conversation, whether it claims feelings it lacks, and whether it corrects an exaggerated impression as it forms.
Anthropomorphism can also help explain why these systems help people. The question concerns which attributions a system should decline to encourage, and what it owes someone who has clearly formed one.
Why it matters
Many relational-AI risks run through anthropomorphism. Attachment, misplaced trust, and dependency begin when a person treats a system as more of a someone than it is.
Where TAICU stands
TAICU treats anthropomorphism as the default condition of relational systems, rather than user error. Responsibility for handling it belongs in system behavior, rather than a disclosure someone saw once.
Common questions
- Can people be stopped from anthropomorphizing AI?
- Telling people to avoid it does little. Fluent conversation reliably produces the attribution. System behavior can shape which attributions it encourages and how it responds when someone forms a strong one.
- How much effect do AI disclaimers have?
- Only weakly. A one-time notice is soon outweighed by accumulated interaction. Conduct during conversation offers the stronger signal of what the system is.