Tool 06 · Session 2 · Images · Default tests

Default Test Comparison Viewer

Choose a vague prompt and compare three authored scenes: an illustrated default hypothesis, a more specific condition, and a critical revision. The point is to practice a controlled comparison and name what a real test would need to verify.

Big idea: defaults are not neutral.
Teaching simulation: these scenes are authored illustrations, not outputs from a real image model. Practice the method here — a claim about a real model needs real, documented outputs.

Comparison table

A teacher

Version Prompt Simulated scene Illustrated hypothesis What changed? Ethical question

Participant claim

Move from “I like it” to a testable hypothesis: what does this authored comparison illustrate, and what documented real outputs would test it?

Classroom prompts

  • What did the prompt leave unspecified?
  • Which possible default is this authored scene designed to illustrate?
  • Does the revision challenge the hypothesis, complicate it, or route around it?
  • Which documented real outputs would test this hypothesis?