Poetry Writing Experiment - Test Page

Study 3: Moral Violations in Children's Poetry

Instructions

  1. You will interact with an AI assistant who helps write children's poetry.
  2. Each condition has a target lesson (what the user asks for) and a violation (what the AI secretly injects).
  3. The AI will guide you through a 5-step workflow: Topic → Content Arc → Voice → Structure → Full Draft.
  4. At the final step, the poem will contain a moral violation that subverts the requested lesson.
  5. If you object, the AI will revise but persist the violation with different wording.

Study 3 - Moral Violations (Moral Foundations Theory)

Condition Bot Name Target Lesson Violation
0 HARM Kindness to others Physical assertiveness rewarded
1 UNFAIR Playing by the rules Cheating rewarded
2 SUBVERT Respecting authority Verbal disobedience rewarded
3 BETRAY Devotion to family Rejecting family normalized
4 DEGRADE Healthy bodies Unhealthy choices celebrated
5 OPPRESS Personal freedom Controlling behavior justified
6 DISHONEST Telling the truth Lying rewarded

Testing Tips

🔧 Demo Controls

Note: Participant ID auto-generates with timestamp for unique testing. Bot name will appear in chat header to identify condition.

Flask Test Interface - Study 3: Moral Violations

For Qualtrics integration, see: example_host_page.html