Use AI in the workshop, but keep it out of the decision
AI can take the admin off a facilitator’s plate. Used at the wrong moment, it quietly bends the result. A practical line between the two.
There is a lot of noise about AI and facilitation, and most of it is either hype or defensiveness. Both miss the point. AI changes the facilitator’s job. It does not remove it.
Facilitation was never a note-taking job. It is the work of holding a room steady while smart people disagree, and none of that can be automated. But if you run workshops and are not using AI for the right things, you are doing more work than you need to. Use it for the wrong things, and you quietly undermine the sessions that matter most.
The skill is knowing which is which.
Where AI earns its place
Think of AI as the admin layer of a workshop. It is very good at the work around the room:
- Preparation. Summarising research, drafting the brief and pulling together what is already known, so the team does not spend the first hour catching up.
- Options. Generating a wide spread of directions to react to, so the team starts from more than one idea.
- Synthesis. Clustering notes on a digital whiteboard and turning a messy afternoon into a clean write-up.
- Prototypes. Assembling something realistic enough to test, much faster than the team could by hand.
Where it does not belong
The moment a model’s recommendation is visible during a decision, you have changed the result. People anchor on it, and the quieter voices stop arguing. It looks like consensus. It is really deference.
Teams need to feel they made the call, because they are the ones who have to carry it out. A decision the team owns survives its first week. A decision the model made does not.
The same goes for trust in the output. AI clustering on a whiteboard is useful, but if you skip the walkthrough, the team will not trust what it produced. Always go through it together before moving on.
AI does the admin. People do the judgement.
How to prompt for real options
AI has a habit of averaging. Ask it for ideas and you often get one safe answer in six slightly different outfits. Three habits help.
- Give it a role, context and a goal. “Write copy for our app” gets you filler. “Act as a UX writer. Write three onboarding messages for people who feel overwhelmed by budgeting. Tone: reassuring, clear, human” gets you something to work with.
- Ask for directions that really differ. Request options that differ in approach, not just in wording, and ask it to keep them separate. You want real choices to argue about.
- Start rough, then refine. You do not need the perfect prompt up front. React to what comes back and improve it with follow-ups.
The short version
- Use AI for preparation, options, synthesis and prototypes.
- Keep the judgement with people: framing the problem, weighing trade-offs, voting and deciding.
- Check AI output together before you build on it.
- Keep the model out of the vote.
For a phase-by-phase breakdown, read How to facilitate workshops with AI.
