The Cybernetic Teammate: How Generative AI Reshapes Teamwork, Expertise, and Collaboration

A field experiment with 776 Procter & Gamble professionals reveals that AI doesn’t just boost productivity—it replicates core benefits of human teamwork, including performance gains, expertise sharing, and positive social experiences.

Why This Research Matters

Organizations justify teamwork on three pillars: teams outperform individuals, cross-functional collaboration integrates expertise, and human connection motivates workers. Generative AI challenges all three assumptions. Researchers from Harvard, Wharton, and P&G designed a pre-registered field experiment to test whether AI can substitute for—or enhance—each pillar.

The Experiment

776 P&G professionals worked on real product innovation challenges across four conditions:

  • Individual without AI (control)
  • Two-person team (R&D + Commercial) without AI
  • Individual with AI
  • Two-person team with AI

Expert evaluators scored each submission blind to condition on quality, novelty, and feasibility. Participants also completed pre- and post-task emotional surveys.

Three Key Findings

1. AI Closes the Performance Gap Between Individuals and Teams

Teams without AI outperformed solo workers by 0.24 standard deviations—confirming that traditional teamwork works. But individuals with AI outperformed the control by 0.37 standard deviations, statistically matching team performance. AI also reduced task time by 16% for individuals and produced substantially longer, more detailed outputs.

For workers whose jobs don’t normally include product development, the effect was especially striking: AI-enabled non-core workers matched the output quality of core-job teams without AI. AI democratized expertise that previously required a more experienced collaborator.

2. AI Dissolves Functional Silos

Without AI, R&D professionals proposed technically oriented solutions while Commercial professionals favored market-facing ideas—a classic silo effect. With AI, both groups produced balanced solutions spanning the technical-commercial spectrum. The distinction between their outputs disappeared entirely.

AI-enabled teams also showed a more uniform distribution of solution types, shifting from a bimodal pattern (dominated by whichever team member held more influence) to a unimodal one. AI reduced dominance effects and balanced contributions across functions.

3. AI Improves Emotional Experiences at Work

Contrary to fears about AI alienating workers, participants using AI reported significantly higher positive emotions (enthusiasm, energy, excitement) and lower negative emotions (anxiety, frustration, distress) compared to the control group:

  • Individuals with AI: +0.457 SD in positive emotions, −0.233 SD in negative emotions
  • Teams with AI: +0.635 SD in positive emotions, −0.235 SD in negative emotions

Solo workers with AI reported emotional responses matching or exceeding those of workers collaborating in human teams—suggesting AI partially fulfills the social and motivational role of a teammate.

When Human Teams Still Win

One important nuance: AI-augmented human teams were three times more likely to produce top-decile solutions (9.2 percentage points above the control baseline of 5.8%). For organizations pursuing breakthrough innovation rather than average performance, human-AI teams outperform individuals with AI. The choice between efficiency and peak performance depends on organizational goals.

What This Means for Your Organization

Rethink team size and composition. If AI-enabled individuals match traditional team performance on average, smaller, more flexible structures become viable—especially for routine product development tasks.

Train workers to prompt effectively. Participants in this study had limited AI experience, suggesting these results represent a lower bound. As prompting skills develop, benefits will compound.

Invest in cross-functional AI fluency. Because AI breaks down silos, workers who learn to prompt across functional domains—not just their own—will generate more balanced, higher-quality proposals.

Don’t dismiss the emotional dimension. AI adoption programs should help workers recognize and internalize performance improvements. Positive experiences with AI correlate with higher expected future use.

The Bottom Line

Generative AI functions as a genuine cybernetic teammate—not a passive tool. It replicates team performance benefits, bridges expertise gaps across functional boundaries, and generates positive emotional experiences for users. Organizations that treat AI deployment as a structural redesign opportunity, rather than a productivity add-on, will capture the most value from the transition.