Avoid misleading anthropomorphism
User-Centered Design
Design Systems
Human-Centered AI Design

Avoid misleading anthropomorphism

Avoid human-like cues that inflate expectations or suggest intent or emotions.

Transparency
Human-Centered AI Design

AI systems are tools, not people. When we make them seem more human-like than they are, we create false expectations and mislead users about their capabilities and limitations.

Anthropomorphism—attributing human characteristics to non-human entities—can make AI feel more approachable, but it also inflates expectations and obscures the reality that AI systems don't have intentions, emotions, or agency.

When I design AI-powered experiences, I avoid misleading anthropomorphism by being clear about what AI systems are: sophisticated tools that process data and generate outputs, not sentient beings with thoughts and feelings.

Misleading anthropomorphism appears in several forms:

  • Language that suggests the AI has intentions, emotions, or personal preferences
  • Visual design that personifies the AI with human-like avatars or faces
  • Behavior that creates an illusion of agency or independent decision-making
  • Emotional manipulation that simulates empathy or emotional connection

While some anthropomorphic elements can make interfaces more engaging, they become problematic when they:

  • suggest the AI has capabilities it doesn't (like understanding emotions or making independent choices),
  • create false expectations about how the system works,
  • or manipulate users into trusting or engaging with the system inappropriately.

By avoiding misleading anthropomorphism, I help users:

  • form accurate mental models of what AI can and cannot do,
  • maintain appropriate expectations about system behavior,
  • and make informed decisions without being manipulated by false emotional cues.

Illusion of Agency or Responsibility

Avoid creating the impression that AI systems make independent decisions or have agency, when they actually follow programmed rules and user inputs.

off on
Explain that AI actions are based on user inputs, settings, and programmed rules. Create the impression that AI systems make independent decisions or have agency. Make it clear that users control settings and can modify or override AI behavior. Use language that suggests the AI "decided" or "chose" something independently.

Language That Implies Intent or Emotion

Avoid language that suggests the AI has intentions, feelings, preferences, or personal experiences.

off on
Use factual language that describes system behavior and outputs. Use language that suggests the AI has thoughts, feelings, or personal experiences. Describe what the system did or found, not what it "wants" or "feels". Use phrases like "I think", "I feel", "I understand", or "I'm sorry" as if the AI has emotions.

Simulated Empathy and Emotional Manipulation

Avoid simulating emotional understanding or using emotional language to manipulate users into trusting or engaging with AI systems.

off on
Offer help and support in a factual, professional manner. Simulate emotional understanding or create false emotional connections. Focus on providing useful information and actionable solutions. Use emotional language to manipulate users into trusting or sharing information.

Visual Personification (Avatars, Faces, Human Traits)

Avoid visual design that makes AI systems appear human-like, which can create false expectations about capabilities and emotional connection.

off on
Use abstract icons or functional symbols to represent AI systems. Use human-like avatars, faces, or anthropomorphic characters that suggest the AI is a person. Clearly label AI systems with descriptive terms like "AI Assistant" or "Chatbot". Give AI systems human names or personal identities that create false expectations.

Why this principle matters

Anthropomorphism can create a false sense of connection and understanding, leading users to overtrust AI systems or misunderstand their limitations.

When anthropomorphism is avoided:

  • users have realistic expectations about AI capabilities,
  • they understand that AI outputs are generated, not "thought" or "felt",
  • and they can make decisions based on accurate understanding rather than emotional manipulation.

Without care to avoid misleading anthropomorphism, users may:

  • attribute human-like understanding or intentions to systems that have neither,
  • overtrust AI recommendations because they feel emotionally connected,
  • or be manipulated into sharing information or making decisions they wouldn't otherwise make.
Adapt to user context and expertise Design for trust and accountability