AI’s ability to synthesize vast quantities of information and ideas at speed is its superpower. But this same strength can derail a workshop if not managed effectively. The traditional facilitation playbook was not designed for a participant that can produce a hundred ideas in the time it takes a human to formulate one. To successfully integrate AI into collaborative sessions, we must rethink our approach to facilitation, adapting our methods to leverage AI's strengths while mitigating its weaknesses.
To understand how AI changes workshop dynamics, it helps to think of a metaphor. Human ideation often resembles a flock of sheep. We tend to follow a leader, are influenced by the first few ideas we hear (a phenomenon known as anchoring bias), and are susceptible to groupthink. Even a room full of brilliant people can end up moving as a single unit, hesitant to stray too far from the initial path.
AI, on the other hand, is more like a colony of rabbits. It has no ego, no fear of judgment, and no cognitive biases. It can explore dozens of divergent paths simultaneously, generating a vast and varied set of ideas without getting tired or fixated. Research has shown that this human-AI collaboration can lead to significantly improved outcomes when structured correctly (1).
This difference presents a new challenge for facilitators. A small group of people armed with AI can generate an overwhelming volume of output in minutes. The classic “diverge-then-converge” model of brainstorming, where everyone puts their ideas on a wall and then groups them, breaks down when faced with hundreds of AI-generated concepts. The convergence phase becomes an exercise in frustration.
To harness the power of AI without drowning in its output, we need to add a new step to the classic ideation process:
1. Diverge with AI: Give participants time to use AI tools to generate as many ideas as possible for a given problem. Encourage them to experiment with different prompts and push the boundaries of the AI’s creativity.
2. Filter with AI: This is the crucial new step. Before sharing the raw output with the group, have participants use the AI to narrow down its own ideas. This is done by providing the AI with a clear set of evaluation criteria and asking it to rank or score its own suggestions.
3. Converge as a Group: Now, instead of starting with a hundred raw ideas, the group begins the convergence process with a smaller, pre-filtered set of the most promising concepts. This allows for a much more focused and productive discussion.
The success of the “Filter with AI” step depends entirely on the quality of the evaluation criteria you provide. You can either define these criteria beforehand or, even better, co-create them with the workshop participants. A common and effective framework is to use a scorecard based on three key dimensions:
- Desirability: How much will customers want or need this solution? Does it solve a real problem for them?
- Viability: Is this solution sustainable for the business? Does it align with our strategic goals and financial constraints?
- Feasibility: Can we technically build this solution with the resources and expertise we have?
By instructing the AI to rate each of its ideas against these criteria, you transform it from a simple idea generator into a preliminary analysis tool. The AI’s ratings provide a structured starting point for the group’s discussion, but it’s important to remember that they are just that—a starting point. The ultimate decision-making power must remain with the human participants, who bring the contextual understanding and strategic judgment that AI lacks (2).
Beyond the modified structure, there are several practical considerations for running a successful AI-enhanced workshop:
Plan for More Time, Not Less: It’s a common misconception that because AI is fast, AI-assisted workshops will be shorter. In reality, participants need more time to write thoughtful prompts, read and evaluate the AI’s output, and iterate on their ideas. Plan for longer individual working sessions than you would in a traditional workshop.
Encourage Collaborative Prompting: While individual exploration is valuable, there is immense power in having small groups of 2-3 people collaborate on writing prompts. The discussion they have while crafting the prompt helps to align their thinking and prepares them to more effectively evaluate the AI’s output.
Embrace Digital Documentation: While physical sticky notes have their charm, AI-enhanced workshops are best run with digital tools. Participants will be copying and pasting large amounts of text, which is cumbersome with physical materials. A shared digital whiteboard or document is essential for capturing and organizing the rich output from these sessions.
The integration of AI into our collaborative processes is still in its early days, and the research on what makes for the most effective human-AI teams is constantly evolving
(3). Some studies show that human-AI collaboration can lead to breakthroughs, while others caution that it can sometimes be less effective than the best human or AI working alone (4).
The key takeaway for CX leaders is that AI is not a magic bullet. It is a powerful tool that, like any tool, requires skill and a thoughtful process to be used effectively. By adapting our facilitation techniques and embracing our new role as the conductors of a human-AI orchestra, we can unlock new levels of creativity and accelerate our journey towards creating truly customer-centric experiences.
(1) Przegalinska, A., Ciechanowski, L., Magnuski, M., & Gloor, P. (2025). Collaborative AI in the workplace: Enhancing human-machine synergy for improved organizational outcomes. Computers in Human Behavior, 152, 108041.
(2) Papamichail, K. N., Alves, G., French, S., Yang, J. B., & Snowdon, R. (2007). Facilitation practices in decision workshops. Journal of the Operational Research Society, 58(5), 614-632.
(3) Xu, J. Q., Liu, Y., & Li, Y. (2025). How the Human–Artificial Intelligence (AI) Collaboration Reduces Cyberloafing: The Role of AI Identity and Openness to Experience. Journal of Business Ethics, 193(1), 1-19.
(4) Vaccaro, M., Weller, A., & Shpitser, I. (2024). When combinations of humans and AI are useful. Nature Human Behaviour, 8(12), 2285-2297.
