Compare human and AI contributions in a three-hour design sprint
Alternate human reasoning, AI support, and explicit comparison from empathy through ideation and rapid prototyping in one compact workshop.
Why
Why it earns its place
Side-by-side contribution logs make acceleration, bias, and weak assumptions discussable. The sprint teaches judgement rather than presenting AI as a universal shortcut.
How
Run it
Set a fair challenge
Provide one non-sensitive brief, evidence pack, timebox, and output criteria. Explain that the published source is a programme, not an evaluated protocol.
Alternate, do not hand over
In empathy, ideation, and prototyping, complete a human move first, request one bounded AI contribution, then compare.
Inspect bias and limits
Mark unsupported empathy claims, repetitive ideas, prototype failures, and what the tool made faster or harder.
Decide what transfers
Teams keep, revise, or reject each AI contribution and write one phase-specific rule for future work.
Ready to paste
Prompt
Support a three-hour design sprint without taking over it.
Challenge: [challenge]
Checked evidence pack with IDs: [paste]
Current phase: [empathy, ideation, or rapid prototyping]
Human work already completed: [paste]
Bounded contribution requested: [one task]
Output criteria: [criteria]
Return the contribution plus: evidence IDs used | assumptions | likely bias or failure | what remains human | quickest verification. Do not invent people, needs, research, or test results. If the evidence does not support the request, say so. Do not decide what advances; wait for the team to compare your contribution with its human work.- What did your contribution make less visible than our human version?
The AIdea layer
The designer’s call
- Human core
- Participants interpret evidence, originate direction, compare contributions, test prototypes, and write the transfer rule.
- The trade
- Fast experiential comparison ↔ rushed phase work. Fix it with one bounded contribution and fixed criteria per phase.
- Skip AI when
- Skip the sprint when participants lack basic prompting practice, a safe evidence pack, or enough time to compare and debrief.
Check before use
Watch out
Keep personal research out of general tools. Three hours encourages shallow acceptance; require an evidence ID, failure note, and human keep/revise/reject decision for every AI contribution.
Source of truth
Evidence
Centrale Pédagogique des IUT publishes a three-hour advanced workshop description; it is a planned programme, not a post-event report or outcomes study.
Centrale Pédagogique des IUT · Enseigner avec l'IA : nouveaux outils et nouvelles démarches