Human × AI × Human

The opportunity is augmentation at the group level

Most discussion about AI focuses on the relationship between one human and one machine: writing faster, analysing faster, automating a task. A more interesting opportunity appears when AI sits inside collaboration between humans.

A group generates far more information than it can consciously process. Conversations happen in parallel. Ideas recur in different language. Weak signals appear at the edge of the room. AI can help capture, organise and synthesise that collective input so the group can see more of itself.

AI can expand the sensing layer

Imagine a workshop with ten tables discussing the same complex problem. A human facilitator can visit each table, but cannot remember every contribution or compare all conversations in real time. With consent and appropriate privacy safeguards, transcripts can be analysed to surface common themes, unusual ideas, tensions, unanswered questions and points of convergence across the room.

The technology does not make the collaboration good. It increases what the collaboration can sense.

Humans provide meaning and judgement

AI does not know which tension matters politically, which idea carries community history, whether a sarcastic remark should be interpreted literally, or which trade-off is ethically acceptable. Those judgements remain human and contextual.

The useful design is therefore not Human versus AI. It is Human × AI × Human: people create the conversation, AI helps reveal patterns, and people interpret those patterns together.

“The best use of AI in collaboration is not replacing the room. It is helping the room see itself.”

ROCeteer

Design trust before technology

Any AI-enabled collaboration needs explicit choices about consent, attribution, retention, access and use. Participants should understand what is being captured, what will be kept, what will be discarded and how the output will influence decisions.

When those choices are vague, the technology can reduce the openness it was meant to support. Trust is part of the architecture.

Use AI to strengthen the human loop

The best applications bring insight back into the room: “Here are three themes appearing across tables.” “These two groups reached the same conclusion from different starting points.” “This minority view has appeared four times but has not reached the plenary.”

That creates a faster learning loop between contribution, synthesis and collective sensemaking. AI becomes collaboration infrastructure — useful precisely because it helps humans connect better with one another.