# AI and Teams =? > Source: > Published: 2026-09-27 06:56:45+00:00 A new team member shows up one day. It’s the chatty, diligent type, always cheerful and eager to help. Works around the clock. Doesn’t eat lunch. All business, but after “thought for 42s”, sometimes hallucinates and goes off on a tangent. A question I’ve been pondering: what happens to our team when we introduce these exemplary AI teammates? How will the team dynamics change? Fortunately, we have scholars ruminating on the same subject such as [New York University professor J.P. Eggers and team](https://hbr.org/2026/08/ai-makes-building-easy-choosing-what-to-build-is-harder). To conduct research on this exact topic, they organized what they called a “buildathon”, a 6-hour workshop, with NYU students to understand AI-infused team collaboration dynamics. Students were handed problems typically faced by New York City residents such as grocery affordability, bike lane safety and access to childcare. Participants were then split into small groups to tackle the cases. As an aid, they received access to AI tools to help them at every step of the way: from researching the problem to creating solutions using prototypes. And, as a finale, the groups revealed their solutions and prizes were awarded to the most compelling ones. With their clipboards in hand, the NYU researchers observed the teams, interviewed them, and assessed how they made progress and collaborated. They also judged the quality of each team’s solution to evaluate how effectively they used AI. And, what were the findings? Three themes stood out. ### **Theme 1: Parallel work creates problems** “... teams that defaulted to parallel work with individual AI chats more often ran into trouble.” A question to you: what’s the norm when working with AI? If you think about it, it’s a highly individualistic endeavor. You sit with your computer. You write some text. You press enter and you receive results. If you’re in a team setting, you work independently in parallel. The downside, as the researchers found, was that it caused “trouble”. “It was challenging because everybody was using different chats, so [it became an] information overload and waste of time,” This relates to **alignment**. With the speed of AI, we can make tremendous progress quickly by ourselves and then we end up with a solution that no one else buys into. You might have seen this quote floating around LinkedIn from time to time. It springs to mind when thinking about the same subject. “If you want to go fast, go alone; but if you want to go far, go together.” Over the last 20 years (much thanks to concepts such as queueing, WIP limits, and context switching), we know that too much in flight at once creates waste. The upside is that you feel productive using AI by yourself, but the cost is losing a collective sense of direction, a common ground. **By working individually, you accrue a debt to the collective that keeps on mounting**. Ultimately, you need the output of the AI to be harmonized with the rest of the group to create alignment of where you should go next. Before AI, we would have (hopefully) stood in front of a whiteboard discussing and weighing options before jumping into individual work. That part is now pushed to the side. “AI works best when used individually, when one person prompts it. Having everyone in the team be in the same place with the AI made me realize our capacity to use the AI was greater as we could all collaborate with it.” As the researchers found, there’s a big opportunity to improve how we as a team collaborate using AI. Here are some ideas for your team. #### What’s the opportunity? **Prompt together** When you’re *shaping* and you’re at an important inflection point, sit together, prompt together. Create alignment by twisting and turning different paths to take. **Explore different options, but set aside time to align** Once you have a sense of direction, for instance a defined *problem to solve*, you can disseminate into individual pathways. For instance, you can prototype different solutions, but come together as a team to weigh options against each other and have important debates, ultimately to set a new direction. **Define principles on “single player” vs. “multi player” AI usage** As a team, facilitate a session to write down some guiding principles on when you should be “single player” vs. “multi player” when using AI. ### **Theme 2: AI for problem framing vs. solutioning** “Several participants noted that AI is “good at execution but not coming up with an idea” We’ve probably all experienced this: the frustration when the AI comes up with something that feels off, generic, sloppy. On the other hand, in other situations, the output feels like magic, it’s the right fit, it’s like the AI knew exactly what we wanted. Relating to the quote above, are there specific use cases where AI creates good vs poor output? What distinguishes “idea” from “execution”? First, what’s an idea? An idea is an insight, the lightbulb moment, when the puzzle pieces finally “click.” It’s product discovery. You’re figuring out the product to build, before building it. You have clarity on the problem to solve. What I believe the participants experienced when they talked about an “idea” is that AI is not well-equipped to find the right problem. That’s a fundamental human undertaking. It comes from being exposed to a multitude of data sources, weaving and wrangling them together and then precision emerges. You’ve been in these settings when it happens. It’s serendipitous. If we outsource that messy affair, we’ll have subpar output. This belongs to the team. When we’re ready, AI is magical at executing our ideas, visualizing, coding them but only after they are shaped. “The winning teams illustrate how insight-led, collaborative decisions on problem definition led to innovative solutions, with AI providing the execution engine they needed to build a working prototype in a day.” The classic quote often attributed to Albert Einstein is highly relevant. “If I had an hour to solve a problem I’d spend 55 minutes thinking about the problem and 5 minutes thinking about solutions.” As the finding shows, there’s fantastic power in sitting together as a team and defining a focus, a problem to solve. If the problem is clear, the AI has the context it needs to help you explore a multitude of possibilities. #### What’s the opportunity? **Be vigilant about the problem definition** Before leveraging AI to come up with solutions, ensure that your team is aligned on a clear problem to solve, neither too big or nor too small. This is the rich context the AI needs to help you come up with solutions. **Lean forward vs. lean back AI** *Lean forward* means you’re actively driving and AI supports you. *Lean back* means you let AI do more of the work. When you’re defining the problem, you need to lean forward while you can lean back more in solutioning. As a team, take time to discuss the cases where you should be “lean forward” versus “lean back” in your AI usage. ### **Theme 3: Diversity of thought is key to unlocking innovation** “The teams that performed best weren’t the most technical. They brought different perspectives to understanding the problem. They had someone who understood the policy landscape, someone who’d experienced the problem personally, someone who could think through business models.” Throughout my career, I have been a huge proponent of diversity of thought. The quote above shows the potential. If we have people in the room who look at the problem from different angles, we’ll unlock innovation. There’s even science backing this. Research conducted by Juliet Bourke shows that **diversity of thinking in teams enhances [innovation by 20 percent](https://aurora50.com/the-business-case-for-diversity-and-inclusion-in-numbers/#bourke)**. The norm of the last two decades has been to shape organizations around small durable product teams consisting of a PM, a designer, and a few engineers. A potential trap of such a model is that the team becomes insular, shielding itself from the “stakeholders” on the outside. But maybe the stakeholders, the experts, are the missing piece when you are discovering something completely new. Because of AI, the game has changed. If AI handles execution, the onus is on understanding the problem, which is where outside expertise can unlock innovation. And, now that the technical barrier is being removed, that lawyer you invited to a workshop can quickly prototype a solution to showcase their unique perspective. There’s huge untapped potential in such a collaboration. #### What’s the opportunity? **Co-create with experts** When you’re shaping the problem or coming up with ideas, bring experts (for instance lawyers, marketers, sales people and so on) into the room who can add important perspectives. Think outside your team boundaries. Set aside time for hackdays and design sprints and other activities where you can involve specialists from the outside to ideate and prototype. **Revisit norms of team structures** Question the “truth” that a product team should only have members from product, design and engineering. Given the context of the team, consider staffing more permanent experts into the team structure, for instance if the domain is compliance-heavy or has a strong marketing focus. ### Conclusion AI + Teams = ? These cheerful AI partners work like magic, but the research from NYU shows the real magic depends on the team: a team that *works together*, *shapes the problem* and has *diversity of thought*. Consider these three important lessons: - As a team, take a step back and understand when you should work by yourself vs. as a team. Acknowledge that there’s a collective debt, misalignment, that mounts when you work individually using AI - At the onset, come together to define the problem to solve to create precision before jumping into solutions - Bring in expertise from outside your team to add diverse perspectives Harness what we’re best at as humans and connect human intelligence and ingenuity with AI’s power of execution. So, the equation becomes: **AI + Aligned, diverse teams = Innovation**