0–5 minutes · Start with a useful signal
Each person names one time a phone helped them: a message, a map, a lesson, a creative tool. Then name a time an attractive signal consumed attention without delivering much. Keep the examples specific; nobody needs to disclose a private experience.
5–12 minutes · Separate cue from consequence
Choose one example. Put three headings on a sheet of paper: what attracted us, what happened, and what we still do not know. Which outcome could be measured? Would more clicks necessarily mean a better result?
12–20 minutes · Challenge an analogy
Compare a hatchling drawn toward a shore lamp with a person following a notification. What makes the comparison useful? Where does it fail? Humans interpret, negotiate, remember, and choose in ways this turtle game cannot represent. Find one claim in the research guide and ask whether it actually tests the human behavior under discussion.
20–27 minutes · Examine AI authorship
Would the same paragraph deserve different scrutiny if written by a human or generated by AI? What matters more: who drafted it, whether its sources support it, whether its uncertainty is clear, or how someone takes responsibility for corrections? Read the book’s authorship disclosure before deciding.
27–30 minutes · Leave with a small test
Choose one reversible change for a specific purpose: a quieter evening, a more focused lesson, or a source check on an AI answer. Say what result would support the change and what would make you reconsider. Agreement is optional; an inspectable question is enough.
Invite disagreement without making anyone defend an identity or diagnosis. Separate personal experience from population claims. Let people pass. These prompts support conversation; they are not therapy, a school policy, or a measure of someone’s well-being.
