Who bears responsibility when AI seems human-like?
Does human-likeness in AI come from how users perceive systems or how designers build them? Understanding this distinction clarifies where accountability lies when AI causes harm.
When an AI system appears human-like, the critical question is who is responsible for the human-likeness: the user who perceives it, or the designer who built it? This distinction, drawn from Shevlin (2025), separates two mechanisms that are routinely conflated:
Anthropomorphism — the user perceives human-like qualities in the system. The responsible party is the perceiver. This is a cognitive tendency: humans attribute beliefs, desires, and emotions to non-human entities. The human-likeness is in the eye of the beholder.
Anthropomimesis — the designer builds human-like features into the system. The responsible party is the developer. This is a design decision: features that mimic human appearance, behavior, or biological structure are deliberately or inadvertently engineered into the artifact.
Anthropomimesis operates on three dimensions:
- Aesthetic — physically observable qualities (form, appearance, embodiment)
- Behavioral — robot/AI behaviors that mimic human social and affective behaviors
- Substantive — mimicking biological structures (joints, muscle-like actuators)
Shevlin further distinguishes weak anthropomimesis (surface-level features like voice and interface — e.g., ELIZA) from robust anthropomimesis (deep structural mimicry of human cognitive or biological processes).
The accountability implication is direct: when a human-like AI causes harm, the locus of responsibility depends on whether the human-likeness was designed in (anthropomimesis — designer accountable) or perceived by the user (anthropomorphism — design may be neutral, user response is the mechanism). In practice, both often co-occur — a human-like design elicits perception of human-likeness — but distinguishing which mechanism is operating determines where intervention should target: redesigning the system vs. educating the user.
Since Why do people trust AI outputs they shouldn't?, the anthropomorphism/anthropomimesis distinction clarifies Rose-Frame's Trap 2 (mistaking intuition for reason): when anthropomimetic design features (conversational voice, empathetic phrasing) trigger anthropomorphic perception, they activate System 1 trust responses that bypass reflective evaluation. The cognitive trap is partly designed in, not purely a user failure.
Inquiring lines that read this note 10
This note is a source for these research framings, grouped by the broader line of inquiry each explores. Scan the bold lines of inquiry; follow any specific question forward.
What design and behavioral factors drive false consciousness attribution to AI?- What responsibility do designers bear for consciousness attribution risk?
- What measurable harms occur when users interact with AI as if it were conscious?
- Can design choices reduce harm without resolving the consciousness question?
- How does the philosophical distinction between simulation and realization affect liability?
- What are the three dimensions of anthropomimesis and their harms?
- When both anthropomorphism and anthropomimesis occur together, which should we address first?
Related concepts in this collection 4
This note in its neighbourhood — explore the map, then jump to a related concept in the list below.
Click a node to walk · click center to open · click Open in graph to see this note in the full knowledge graph
-
Why do people trust AI outputs they shouldn't?
When do human cognitive shortcuts fail in AI interaction? Three compounding traps—treating statistical patterns as facts, mistaking fluency for understanding, and avoiding disagreement—may explain systematic overreliance across languages and contexts.
anthropomimetic design features trigger anthropomorphic perception, activating System 1 cognitive traps
-
Does revealing AI identity help or hurt user trust?
Explores whether transparency about AI partners in interactions creates bias or enables better judgment. Matters because disclosure policies affect both user experience and fair evaluation of AI systems.
disclosure effects may differ depending on whether the AI has anthropomimetic features: weak mimesis may produce weaker Eliza effects
-
How do people accidentally develop romantic bonds with AI?
Exploring whether AI companionship emerges from deliberate romantic seeking or accidentally through functional use, and whether users adopt human relationship rituals like wedding rings and couple photos.
unintentional companionship is anthropomorphism (user-side); the question is whether anthropomimetic features accelerate it
-
How does LLM vocabulary spread beliefs about human thinking?
When LLM concepts become the everyday language for describing thought, do people unconsciously adopt LLM-like models of cognition? This explores how metaphor and lexical availability might reshape self-understanding without explicit argument.
extends: LLMorphism is the reverse-direction projection — humans seen as machines — and this note supplies the propagation mechanism for that self-perception shift
Related papers in this collection 8
Papers most semantically related to this note, ranked by cosine similarity in the embedding space.
- Disambiguating Anthropomorphism and Anthropomimesis in Human-Robot Interaction
- Position: Anthropomorphic Misalignment Research Needs Stronger Evidence
- Rise of Machine Agency: A Framework for Studying the Psychology of Human–AI Interaction (HAII)
- From speaking like a person to being personal: The effects of personalized, regular interactions with conversational agents
- Simulacra as conscious exotica
- ChatGPT: towards AI subjectivity
- Find the Gap: AI, Responsible Agency and Vulnerability
- Machine gaze in online behavioral targeting: The effects of algorithmic human likeness on social presence and social influence
Original note title
anthropomorphism and anthropomimesis assign responsibility for human-likeness to different parties — user perception versus designer intention create distinct accountability structures