AI as a Teammate? A Groundbreaking Study Says Yes—And It’s Surprisingly Human
If AI agents are going to fit into our workplace, they need to do more than just perform—they need to collaborate, communicate, and connect —like good teammates.
A recent post by Ethan Mollick, The Cybernetic Teammate, highlighted a fascinating new study that shows this might already be happening.
Mollick, a Wharton professor who’s been leading the charge in exploring how generative AI works in the real world, walks us through a peer-reviewed field experiment that took place inside Procter & Gamble. And it’s not just academic theory—this was a real-world test with 776 professionals, solving actual innovation challenges, with and without the help of AI.
We looked at the research through our own Good Agents lens—specifically the “Human-focused” part of our FARSIGHT framework. We asked: Is this AI actually serving the human? Is it adapting, supporting, and collaborating in a way that feels natural and useful?
Here’s what stood out:
👥 From “Tool” to “Teammate”
One of the big takeaways: individuals using AI performed as well as teams without AI. That’s massive. It means AI isn’t just helping you do things—it’s helping you think better, faster, and more collaboratively, even when you're working alone.
And the kicker? People didn’t just use AI to grind through tasks. They described it more like a teammate—offering feedback, encouraging new ideas, and helping them look at problems from more than one angle. That’s right in line with what we mean when we say a Good Agent should act like a collaborative equal, not just a subordinate tool.
🧠 Smarter Together: Bridging the Expertise Gap
Another brilliant finding: AI helped professionals step outside their silos. R&D folks typically lean technical, while commercial folks focus on the market. But with AI? Everyone started generating well-rounded, balanced proposals. That tells us the AI was adapting to different users’ domain knowledge—helping them understand and speak each other’s language.
That’s a clear tick on our checklist item: “Adapt to different levels of technical or domain expertise.”
😊 And Yes, It Made People Feel Better
This one surprised even the researchers. Participants using the AI reported more positive emotional responses. They liked the experience. They felt supported. That simple, conversational interface wasn’t just functional—it was motivational. The AI wasn’t just making people more productive; it was making them feel better while they worked.
That’s what Human-focused AI should do. It should pick up on emotional tone, reduce friction, and reinforce confidence. It shouldn’t just do the job—it should do it in a way that feels good.
So, What Does This Mean?
This research suggests that AI, when used right, doesn’t just make teams more efficient—it can redefine what teamwork looks like. It can help us collaborate across silos, think more broadly, and maybe even enjoy our work a little more.
📖 Want to dig in? The original study: The Cybernetic Teammate
🧠 And Ethan Mollick’s summary: One Useful Thing