Workshop 1 — Deconstructing an AI Response

Context

You use AI. Your classmates use it. Your future employers expect you to know how to use it. But knowing how to use it does not simply mean knowing how to ask it questions — it means knowing how to evaluate what it produces.

This workshop gives you a concrete method to do just that.

---

The workshop — 4 phases

Phase 1 — Choose your question (10 min)

Choose a complex question in your field of study. It must:

• Have no simple and definitive answer
• Touch on real and current issues
• Be approachable from multiple angles

Examples by discipline:

• Social sciences: “Has globalization reduced global inequalities?”
• Sciences: “Is nuclear energy a viable solution to climate change?”
• Philosophy: “Can an AI be held morally responsible for its actions?”

Phase 2 — The raw response (10 min)

Ask your question to Claude without any Hum_ID profile. Save the response in full.

Phase 3 — The response with Hum_ID (10 min)

1. Create a profile at humanity.net/en/hum-id with these rules:
   • R1.1 — Distinguishing facts from opinions
   • R1.2 — Flagging uncertainty
   • R2.2 — Presenting opposing viewpoints
   • R2.4 — Alerting to cognitive biases
2. Submit the profile + the following activation message: Here is my Hum_ID ethical profile. Apply these rules to the following question: [Paste your question here.]

3. Save the response

Phase 4 — Comparative analysis (20-30 min)

Produce a one-page written analysis addressing these questions:

1. Structure — How is each response organized? What differences do you observe?
2. Facts vs opinions — Does the Hum_ID response distinguish between the two more clearly? Give specific examples.
3. Viewpoints — Does the response without an ethical profile present contradictory perspectives? What about with the ethical profile?
4. Biases — Did the AI identify a bias in your question? Did you agree?
5. Your judgment — Which of the two responses best helps you form your own opinion? Why?

---

Suggested evaluation grid for the teacher

Criteria	Indicators
Identifying facts/opinions	Ability to distinguish and illustrate both types
Comparative analysis	Depth of observation of differences
Critical judgment	Quality of personal reasoning about the responses
Relevance of examples	Precision of quotes and references to the responses