The Cybernetic Teammate: What a P&G Field Experiment Reveals About the Future of Collaboration
At The Change Lab, we often talk about the intersection of behavior, systems, and technology. A new field experiment by Procter & Gamble, in partnership with researchers from Harvard and Wharton, offers a compelling data point at that crossroads. Here is the Article
The study asked a deceptively simple question: Can generative AI act like a teammate? The researchers tested 776 P&G professionals—Commercial and R&D—on real innovation challenges. Some worked solo, others in teams. Some had access to GPT-4, others didn’t. The findings were striking:
Individuals using AI matched the performance of human teams working without it.
AI flattened functional silos, enabling Commercial and R&D participants to produce balanced ideas, regardless of their background.
Participants using AI reported more positive emotional experiences—more enthusiasm, less frustration—echoing the social benefits we usually associate with human collaboration.
This isn’t just about productivity. It’s about how we work, who gets to contribute, and how we design systems that enable better thinking. For organizational change practitioners, the implications are profound.
Implications for Change Management
AI as a Change Agent, Not Just a Tool
This study reinforces a key idea we’ve been seeing in Copilot rollouts: AI isn’t just a new application to adopt—it alters the nature of collaboration. It can replace some benefits of teaming, extend capabilities into new domains, and shift how people feel about their work. That calls for change strategies that go beyond training. We need narratives and pilot designs that surface new ways of working, not just new tools.
Breaking Expertise Silos by Design
In behavioral economics, functional silos often act like default heuristics—we fall back on our domain knowledge when under time pressure. This study shows that AI can disrupt that pattern. It acts as a cognitive bridge, helping people navigate commercial and technical domains with equal fluency. For OCM, this raises the question: How do we enable more fluid, boundary-crossing work? And how do we reinforce that behavior before AI becomes just another invisible layer?
Emotions Matter—Especially During Adoption
One of the most underreported findings in tech adoption is the emotional component. This study quantified it. Participants using AI felt more energized and less anxious. That’s a shift worth capturing. In our change plans, we often talk about Awareness and Desire (ADKAR), but here’s a reminder that emotional feedback loops are part of how people build confidence and change behavior. AI isn’t just increasing what people can do—it’s changing how they feel while doing it.
Where This Fits
At a time when many organizations are still framing AI as a bolt-on capability, this research points toward a deeper integration—one where AI becomes part of the social fabric of work. For change leaders, this means treating AI not just as a deployment challenge, but as a redesign moment. A chance to rewire how people, roles, and feedback loops come together.
The term they use is “cybernetic teammate.” It’s worth considering whether our organizations are ready to make space for that kind of partner.
