Design to prevent false human attribution.
Actively discourage mental models that treat the system as a person.
Contents 9. AI-Mediated Interaction 10. Non-Anthropomorphism
How to prevent false human attribution.
First, ask yourself
“Does your system present itself honestly to users as a tool, not a human agent?”
Mission statement
Ensure users form accurate mental models by presenting the system as a tool while avoiding human-like cues.
Key heuristics Non-Anthropomorphism
- System design must prevent false human attribution.
- Cues must not imply human emotion, empathy, or intent.
- Language must reflect functional capability, not awareness or desire.
- Outputs must be framed as computational processes, not deliberate acts.
- System constraints, uncertainties, and boundaries must be clearly communicated.
- Responsibility must not shift to a system persona.
- Metaphors must avoid implying thought, belief, or motive.
- Polished interaction must not imply comprehension or epistemic authority.
- Human-like interaction patterns must be reduced in high-risk contexts.
Executive brief
AI systems must avoid human-like framing or language that may imply sentient comprehension.
Core questions Non-Anthropomorphism
“Does the system clearly communicate what it is?”
“Does it prevent users from attributing human intent, understanding, or agency?”
Focus areas
Mental-model first
“What does the interface teach users about the system?”
“Is the interface encouraging accurate understanding or social projection?”
Responsibility-aware
“Does personification obscure accountability?”
“Are decisions framed as system behavior rather than human judgment?”
Interaction-aware
“Is system feedback interpreted without anthropomorphizing?”
“Does the UI reinforce correct causal reasoning about the system?”
AI-aware
“Are uncertainty and limits made explicit?”
“Does language exaggerate intelligence, care, or agency?”
UX directives Non-Anthropomorphism
Prevent cues that imply human emotion or intent.
Do not simulate empathy or consciousness to increase engagement.
Use language that reflects functional capability.
Avoid phrasing that implies awareness, desire, or understanding.
Represent system agency truthfully.
Frame outputs as computational processes, not deliberate actions.
Make system limitations explicit.
Clearly communicate constraints, uncertainties, and boundaries.
Preserve human accountability in system framing.
Ensure responsibility is not shifted to the system persona.
Prefer functional metaphors over social metaphors.
Refrain from metaphors that imply thought, belief, or motive.
Separate fluency from epistemic authority.
Prevent polished interaction from implying comprehension.
Limit human-like interaction patterns in high-risk contexts.
Reduce social signaling where overtrust would be harmful.
Executive summary
- Non-Anthropomorphism prevents false human attribution, not expressive interaction.
- It ensures users understand the system as a computational process, not a person with intent or emotion.
- The system must use functional language and avoid cues that imply awareness, belief, or motive.
- Fluency of interaction must never imply comprehension or epistemic authority.
- Limitations, boundaries, and accountability must remain explicit and human-centered.
- Non-Anthropomorphism succeeds when users reason about capability accurately without projecting agency where none exists.
Success indicators
- The system is presented as a tool, not a person.
- Language describes functions rather than thoughts or feelings.
- The system’s limits and uncertainties are clearly communicated.
- Responsibility remains clearly attributed to human actors.
- Polished responses do not imply human understanding or intent.
One-line summary
Before users can understand what the system is, it must not present itself as a human agent.