Compare design work without AI, with AI, and fully AI-supported
Take one challenge through three comparable passes to experience where AI accelerates analysis and where human creativity should lead.
Why
Why it earns its place
A three-pass comparison lets participants inspect AI's contribution directly. A fixed challenge and evidence pack keep the debrief focused on contribution, not tool novelty.
How
Run it
Prepare a fair comparison
Set one challenge, one evidence pack, fixed teams, equal timeboxes, and a common output checklist. Explain that this controlled repetition is an AIdea adaptation.
Work without AI
Complete the agreed design cycle manually. Record time, key decisions, evidence used, and the resulting output.
Add AI as a teammate
Keep people in charge and assign AI bounded contributions. Log every accepted, rejected, and corrected suggestion.
Support the full process, then compare
Allow AI support at each stage while retaining human accountability. Compare evidence fidelity, usefulness, creative quality, effort, and where AI-first or human-first sequencing worked best.
Ready to paste
Prompt
Support one bounded contribution in an AI-assisted pass of this comparison.
Current pass: [AI as teammate or fully AI-supported]
Design challenge: [challenge]
Current stage: [stage]
Evidence pack with IDs: [paste]
Human work already completed: [paste]
Requested contribution: [one analysis or expansion task]
Output criteria: [criteria]
Use only the supplied evidence. Return: contribution | evidence IDs used | assumptions | uncertainty | human checks required. Do not invent users, findings, facts, decisions, or test results. If the evidence is insufficient, say what is missing instead of completing the task. Do not choose the direction; wait for the team to log, compare, accept, reject, or correct your contribution.StretchSwap the order of the AI-assisted passes and discuss whether sequence changed the team’s judgement.
The AIdea layer
The designer’s call
- Human core
- People frame the challenge, judge evidence, originate creative direction, and decide what the comparison means.
- The trade
- Experienced judgement ↔ a long, facilitator-heavy activity. Fix it with one compact challenge and identical criteria across passes.
- Skip AI when
- Skip the comparison when teams cannot hold inputs and timeboxes steady enough to make the debrief meaningful.
Check before use
Watch out
Do not treat the exercise as a productivity race. Verify every analytical claim against the common evidence pack, log corrections, and keep confidential inputs out of unapproved tools.
Source of truth
Evidence
HPI reports one professional’s positive reflection on a three-day workshop; the three modes are source-grounded, but the same-challenge comparison protocol is an untested AIdea adaptation.
HPI d-school · Interview about AI meets Design Thinking