A field experiment with 791 professionals at Procter & Gamble offers a glimpse of what teams may become. People working alone with AI performed about as well on the studied product-innovation tasks as human teams working without AI.
AI also helped people think beyond their usual specialties. Technical professionals produced ideas with more commercial balance. Commercial professionals produced ideas with more technical balance.
That does not mean one employee plus AI can replace every team. It means some of the thinking we once needed a meeting to produce can now happen before the meeting begins.
Imagine sending a proposal through an AI review before the real team sees it. Ask the system to examine the proposal as a technologist, finance leader, customer, regulator, communications director, frontline employee, donor, and skeptic. Let it surface contradictions, weak assumptions, and unanswered questions.
Then bring in the actual specialists.
They no longer have to spend their first hour finding the obvious problems. They can apply what AI does not truly possess: lived experience, institutional memory, political judgment, relationships, and responsibility for the decision.
This can make human collaboration more valuable, not less. AI can widen the range of perspectives available to a team. People still have to decide which perspectives are credible, which tradeoffs are acceptable, and what the organization is willing to do.
Teams also perform social work that fluent language can imitate but not replace. Trust, conflict resolution, mentoring, courage, negotiation, loyalty, and shared responsibility affect what people will support and what an organization can actually execute.
The cybernetic team therefore uses AI to expand and integrate ideas while people authenticate, challenge, negotiate, and decide.
The modern team's value lies less in gathering information and more in combining accountable perspectives.
The Procter & Gamble field experiment involving 776 professionals provides a useful boundary marker. Individuals using AI could match the performance of human teams without AI on the studied product-innovation tasks, and AI helped specialists work beyond their usual functional boundaries. That is a major change in the economics of collaboration.
But it does not follow that teams disappear. Some of what teams once provided was access to distributed knowledge, and AI can partially reproduce that. Other team functions are social and institutional: trust, negotiation, mentorship, conflict, legitimacy, courage, and shared responsibility. Those do not vanish because information is easier to generate.
The cybernetic team therefore uses AI to reduce the cost of perspective while preserving the value of accountable human relationships.
Before the meeting
The cybernetic team changes what should happen before humans gather. AI can compare documents, identify contradictions, generate stakeholder questions, model scenarios, and prepare competing interpretations. That means expensive human meeting time can move away from information retrieval and toward disagreement, negotiation, judgment, and commitment.
This is a subtle but important redesign. The goal is not to simulate a team so that the real team can be eliminated. It is to let machines perform some of the cognitive preparation that previously consumed the meeting. Human collaboration becomes more valuable when people arrive better prepared to use the things only a real group can supply: trust, institutional memory, political judgment, courage, and shared responsibility.
The complementarity test
For every team workflow, ask what the machine makes cheaper and what that should allow humans to do better. If AI shortens research, perhaps humans should spend more time testing assumptions. If it drafts options, perhaps the team should spend more time on tradeoffs. If it summarizes meetings, perhaps leaders should spend more time resolving ambiguity. Productivity is wasted when saved time simply disappears into more volume.
The P&G case and its limit
The Procter & Gamble field experiment is important because it demonstrates genuine complementarity. In product innovation work, individuals using AI could perform at levels comparable to human teams without AI, and AI helped technical and commercial professionals produce more balanced solutions across functional boundaries.
The tempting conclusion is that teams can now be reduced. A better conclusion is that some functions of teams have become cheaper. Access to alternative perspectives, synthesis across specialties, and first-pass ideation can increasingly be supplied by AI. That should change what humans spend meeting time doing.
But the experiment does not show that AI possesses organizational standing. A model cannot make a colleague trust a difficult decision, absorb the career consequences of a failed launch, negotiate a resource conflict, mentor a future leader, or credibly represent a constituency. These are not mystical human qualities. They are functions produced by relationships, incentives, history, and accountability.
The cybernetic team should therefore be designed around complementarity rather than substitution. Let AI lower the cost of information and perspective. Use the recovered human time for challenge, negotiation, commitment, coaching, and decisions whose legitimacy depends on actual people.
The team does not disappear when information gets cheap Many meetings exist because information is fragmented. People gather to report what they know, reconcile documents, generate options, and summarize previous discussions. AI can compress much of that coordination. It can prepare the shared brief before the meeting, surface disagreements in advance, and generate scenarios that once consumed meeting time.
That should reduce the need for some meetings. It should also raise the standard for the meetings that remain. If everyone can arrive informed, the human gathering should be used for what is difficult to automate: negotiating interests, testing commitment, resolving ambiguity, reading hesitation, making tradeoffs, and establishing shared responsibility.
Complementarity is designed, not discovered
The P&G field experiment is instructive because AI did more than make individuals faster. It also helped professionals cross functional boundaries. Technical participants could produce more commercially oriented thinking, and commercial participants could contribute more technically oriented ideas. In the studied task, individuals with AI could reach performance levels comparable to teams without AI.
The tempting conclusion is that AI replaces teams. The more useful conclusion is that AI changes what teams are for. If AI can provide some informational diversity, then human teams can spend more of their scarce time on disagreement, legitimacy, tacit context, and commitment. The value of the team shifts from assembling knowledge to integrating accountable perspectives.
A cybernetic meeting protocol
Before a consequential meeting, use AI to produce a common factual brief, identify unresolved assumptions, simulate stakeholder objections, and list decisions that actually require human authority. During the meeting, prohibit the group from spending most of its time re-summarizing information everyone already has. After the meeting, use AI to capture decisions, owners, evidence needs, and open questions, then have humans correct the record before it becomes institutional memory.