A field experiment with 791 Procter & Gamble professionals gives one of the clearest answers yet to what generative AI does for teamwork. Working on real product-innovation challenges, participants were randomly assigned to work alone or in pairs, with AI or without. People working alone with AI performed as well as two-person teams without it.¹

The second finding may matter more for anyone who runs cross-functional work. Without AI, R&D professionals proposed technical solutions and commercial professionals proposed commercial ones. With AI, both groups produced more balanced ideas, whatever their background. Participants using AI also reported more positive emotions during the task. When the researchers broke the process down, AI mainly raised the quality of the ideas generated, while human judgment kept its value in choosing among them.¹

Who gets the gain

Other field experiments point the same way on output. Across three randomized trials at Microsoft, Accenture and a Fortune 100 company, 4,867 software developers with an AI coding assistant completed 26 percent more tasks. Less experienced developers adopted the tool more and gained more,² a pattern we saw in customer service as well, where the AI boost went to the rookies.

A study from Kenya marks the limit. Researchers gave 640 entrepreneurs access to a GPT-4 business assistant on WhatsApp and compared them with a group that received static training materials. On average, revenues and profits did not change measurably. Underneath the average, businesses that had performed well at the start may have gained more than 15 percent, while low performers did nearly 10 percent worse. Both groups asked similar questions and received similar advice. The difference lay in which advice they chose to act on.³

Taken together, the studies show AI raising output where people can judge what it offers. How far it can stand in for a colleague depends on the task. The P&G result held for a specific innovation task, and our piece on consultants shows how quickly the gain turns once a task crosses the edge of what the tool does well.

By the Numbers

791 P&G professionals: alone with AI matched a team of two without it¹
4,867 developers: 26% more completed tasks with an AI assistant²
640 entrepreneurs in Kenya: high performers may have gained more than 15%, low performers did nearly 10% worse³

Talent Takeaway

When cross-functional meetings keep replaying the same two positions, a small experiment can help: each side drafts its proposal with AI beforehand. The Kenya study adds a condition. People gain from AI advice when they can judge it, so building that judgment belongs in any rollout plan.

¹ Dell'Acqua, F., Ayoubi, C., Lifshitz, H., Sadun, R., Mollick, E., Mollick, L., Han, Y., Goldman, J., Nair, H., Taub, S., & Lakhani, K. R. (2026). The cybernetic teammate: A field experiment on generative AI and teamwork. Organization Science, 37(4), 1217–1242. https://doi.org/10.1287/orsc.2025.20702

² Cui, K. Z., Demirer, M., Jaffe, S., Musolff, L., Peng, S., & Salz, T. (2026). The effects of generative AI on high-skilled work: Evidence from three field experiments with software developers. Management Science. Advance online publication. https://doi.org/10.1287/mnsc.2025.00535

³ Otis, N. G., Clarke, R., Delecourt, S., Holtz, D., & Koning, R. (2026). The uneven impact of generative artificial intelligence on entrepreneurial performance: Evidence from a field experiment in Kenya. Management Science. Advance online publication. https://doi.org/10.1287/mnsc.2024.06909

Further reading: Columbia Business School. (2026, August 20). AI widens the gap between high and low performers. https://business.columbia.edu/insights/ai-widening-gap-between-high-and-low-performers