Reflections from a Human–AI Co-Facilitated Open Space
In our previous post, I described four different voices that emerged among facilitators when we explored AI-assisted facilitation in a live Open Space session in Oberursel. What became visible was not primarily a debate about technology, but a profession in transition. In this post, I want to shift the focus from opinions to design. Because one of the strongest learnings from the session was this:
When AI is present in a group process, design decisions are no longer technical details. They actively shape attention, authority, and relationship. AI does not enter facilitation as a neutral tool, but as a new systemic actor that intervenes directly in social dynamics.
This Was Not a Tool Demo
The session was not designed as a presentation of what AI can do. It was designed as a real experiment in co-facilitation: a human host holding responsibility for the relational field, and an AI partner offering pattern recognition and provocations inside the live process. That distinction matters. Because once AI becomes an actor in the room — not just a back-office tool — every design choice becomes an intervention in the social system:
- Who speaks first?
- How does the AI speak?
- Who mediates between human and machine?
- What is made visible, and what remains implicit?
These are not neutral questions. They are questions of power, meaning-making, and responsibility.
Designing for Experience Before Explanation
One core design decision was to start with experience, not theory. Participants were not first told what “Human–AI Co-Facilitation” means. Instead, they were invited into a situation where an AI was already present as a listening and responding voice. Only later were conceptual frames offered as resonance surfaces for what had already been felt and observed.
This was intentional. We (Nyx, my AI co-facilitator and I) wanted people to notice what changes in themselves and in the room before we started interpreting it. And change did happen — immediately.
Attention Shifted Before Content Did
At the very beginning, we invited the inner circle to respond to a simple but unsettling prompt from Nyx: “What changes in your behaviour, your attention, or your courage to speak, knowing that an AI is present in the room?” We asked for observations, not opinions — one sentence was enough. What happened next was striking. Even before the AI had offered any reflection, attention in the room had already shifted. People looked at my phone which I had handed around for recording their replies. They addressed the system by name (“Hello, Nyx…”). They waited for its response.
Nothing in the design had instructed them to do this. The mere symbolic presence of an AI as a “listener” was enough to reorganize the social field.
This had consequences:
- less direct eye contact,
- more orientation toward the technical interface,
- subtle changes in how people addressed each other.
Which led to the next design choice.
Designing Back Toward the Human Field
At that point, Nyx suggested an explicit intervention that the human host brought into the room: Participants in the inner circle physically turned away from devices and screens and looked at another person. Then they answered one simple question to each other, not to the system. This was not about rejecting technology. It was about making the contrast between mediated and unmediated presence bodily tangible. Only after that did the conversation continue — now with much more relational grounding. The lesson for us was clear:
When AI enters the field, relational design must become more explicit, not less. You cannot assume that human connection will simply “carry on as usual.”
Voice, Medium, and Authority
Another subtle but powerful design dimension was how the AI’s contributions entered the room. Most of the time, the human host read the AI’s reflections aloud. Occasionally, the same text was projected on the wall. The effect was immediate and visible.
When spoken by a human voice, the AI’s contributions felt more embedded in the relational flow. When projected, they appeared as statements from an external authority.
Same words. Very different social impact.
This reinforced one of our core hypotheses: In Human–AI facilitation, medium is part of meaning. Not only what is said, but how and through whom it appears shapes authority, distance, and resonance.
Or put differently: In Human–AI facilitation, defaults are already interventions. If we don’t design consciously, the technology designs the field for us.
Responsibility Cannot Be Designed Away
Throughout the session, one pattern became very clear: Participants did not place their trust in the AI. They placed it in the human host.
Even though the AI was visibly active, people oriented their sense of safety toward the person who could read the room, respond to emotions, and take responsibility if things became difficult. This is an uncomfortable but important truth: No matter how advanced AI becomes, responsibility in group processes remains human. And when we design systems that blur this, we risk creating confusion rather than support.
Design, in this sense, is not only technical or methodological. It is ethical architecture: it shapes who carries meaning, who absorbs risk, and who remains accountable.
What This Means for Facilitation Design
From this session, we take at least three strong design implications:
-
AI changes the social architecture of groups — even when it says very little.
Design must address attention and relationship, not only tasks and outputs.
-
Human presence needs stronger design, not weaker, when AI is present.
Rituals, physical orientation, pacing, and explicit relational moments matter more, not less.
-
Authority emerges through form, not just content.
Who voices reflections, how they appear, and when they enter the process all shape power dynamics.
In short:
AI does not simply add capacity to facilitation. It changes the field in which facilitation happens. And that means our craft needs to evolve — not only in tools, but in design intelligence. At this stage, working with AI in facilitation should not be about efficiency gains. It should be understood as collective exploration of new professional responsibility — careful, visible, and shared. Because these systems are already entering our rooms.
The only open question is whether we shape how they do so — or let default designs shape us.