Treating AI like a person can help only when students know it is role-play
Cluster note
This page explores AI role-play with boundaries. Related: bounded immersion for learning and calibrated anthropomorphism. Bounded immersion is the broader structure; calibrated anthropomorphism focuses specifically on person-like framing.
Definition
It can be useful to give AI a role, persona, or point of view, but students still need to understand that the model is not a human collaborator with understanding, agency, or trustworthiness. This is especially important because blurred lines between reality and AI interaction, unhealthy dependence, and LLM-psychosis-style risks are real safety concerns schools should take seriously.
Current synthesis
Person-like framing can improve AI interaction when the frame is treated as deliberate role-play rather than evidence that the system understands, cares, or has agency. The Power of Treating AI Like a Colleague AI Sycophancy Is Not Always Harmful
Kentz adds that some of the learning power comes from temporarily “committing to the bit,” but that benefit depends on boundaries, explicit exits, and post-use reflection so the simulation does not blur into trust or emotional dependence. The Commitment Paradox To Learn From
The Gatsby interview example shows a classroom-safe version of this pattern: students can treat the bot as a character interlocutor while staying analytically above it, using contradiction, evasion, and limitation as evidence rather than authority. When Students Interview Jay Gatsby
Plain-language note
This page is also about what researchers may call calibrated anthropomorphism: using human-like roles as a temporary learning scaffold while keeping a clear reality check.
Articles
- The Power of Treating AI Like a Colleague
- AI Sycophancy Is Not Always Harmful
- The Commitment Paradox To Learn From
- When Students Interview Jay Gatsby
Linked claims
- Treating AI like a person can help if students know the limits
- Prompting AI is a literacy practice, not just a technical trick
Relationship to linked claim
This page is the umbrella for calibrated, role-based AI interaction. The linked claim Treating AI like a person can help if students know the limits is the guardrail: person-like framing is useful only when students understand that the persona is a temporary role-play scaffold and not a real collaborator.
Related syntheses
Open questions
- How should this idea be translated into concrete classroom routines, policies, or professional learning?