It’s Hard to Use AI as a Team. These 3 Practices Can Help.

According to a global survey of 500 executives conducted by the Capgemini Research Institute, the active use of AI in team meetings is anticipated to more than triple in the next three years. The executives surveyed said that they expect group use of AI to lead to better meetings that arrive at higher-quality outcomes. Leaders should not let this optimism obscure the challenges ahead. Our research suggests that integrating AI into team settings doesn’t happen naturally, and introducing AI into meetings without laying the proper groundwork can narrow participation, fragment discussions, or shift ownership away from the team.

Fortunately, there is an approach that overcomes these pitfalls: we call it “Human-AI Team Chemistry.” Our research indicates three practices to help build this new capability as AI becomes more embedded in organizations:

1. Engage with AI as a team. Participants should introduce themselves and involve the AI in a collective dialogue so that it addresses the group, considering the various expertise at the table.

2. Leverage AI’s role fluidity. AI should be used not just as a note-taker but as a multi-role team member, deliberately switching roles (such as stakeholder representative, challenger, customer, competitor, etc.) to enrich the team discussion.

3. Maintain collective ownership of the interactions with AI. When team members treat prompting as a collective act, debate alternative directions, and pause to judge AI’s output throughout the meeting, interactions with AI will advance their thinking rather than outsourcing it.

These recommendations emerged from a five-month experiment involving 60 managers from 12 companies across diverse industries. In each organization, a team of three to four managers—all with prior individual experience using generative AI—was tasked with designing a platform-based solution to address a strategic business challenge of comparable scope and complexity. To ensure consistency, all teams followed the same methodology, developed by two of us (Daniel and Tommaso), and used OpenAI’s ChatGPT model. Each team met in person five times, for a total of 30 hours of collaborative work. To understand how teams actually worked with AI and experienced the process, we used three complementary methods: we observed team interactions with the AI in real time, analyzed the complete chat transcripts from each session, and collected post-session surveys to capture participants’ feedback. This combination allowed us to see not just what teams produced, but how they collaborated and where team-AI collaboration thrived or faltered.

Based on this research, we believe that teams that engage in these practices will achieve higher‑quality outcomes and reduce the risk of falling into common AI‑related traps.

Session One: False Starts and Missed Potential

In the first session, the idea of integrating AI into teamwork felt intriguing and new to all participants. But the initial excitement faded quickly, often within the first hour. Teams grew quieter, slipped into a more passive mode, and began to simply watch the screen as the AI generated responses. AI was dominating the conversation, while the team gradually receded into the background. Survey data after the first session confirmed what we had observed: teams reported limited perceived benefits, and overall engagement was low. We were puzzled: the result was the opposite of what we had expected. Rather than amplifying collaboration, AI seemed, at least initially, to dampen it.

Defaulting to individual chat mode

A closer review of chat transcript revealed the roots of the issue. As one of us noted, “If I didn’t know this was a team chat, I would have assumed it was an individual interacting with an AI.” In a human-only team meeting, when a new colleague or consultant joins, everyone introduces themselves, explains their role, and shares context so the newcomer can contribute effectively. When interacting with AI, teams did not follow the same norms. As a result, the AI responded as though it were supporting only the person typing, rather than engaging with the full group or accounting for its dynamics. Lacking awareness of the team’s context, its different roles, expertise, and viewpoints, the AI defaulted to a narrow, individual-focused perspective.

Giving AI a static role

Another interesting pattern emerged: most teams assigned the AI a single, static role—often “researcher” or “subject‑matter expert”—and left it at that for the entire session, interacting with it primarily in answer-seeking mode, treating it as a repository of expertise to query rather than a thinking partner. None experimented with shifting the AI into more abrasive roles, such as critic or skeptical stakeholder, which could have encouraged reflection, constructive debate, or more active challenge. Nor did they ask AI to take different perspectives—a customer, a competitor, an end user—which might have revealed blind spots or hidden assumptions.

Interacting staccato-style

We also noticed that the teams’ input to the AI was typically brief and transactional: “OK, go ahead,” “give me another example,” “this isn’t the right direction.” These quick, minimal requests reveal that the team members rushed through, without articulating their goals or explaining their reasoning to the AI. Furthermore, AI often jumped ahead by proposing unsolicited “next steps” or ready‑made options, nudging the team toward simple one-click confirmation (“OK, Option B”) and steering the conversation before collective alignment had occurred.

The same three pitfalls appeared repeatedly across groups, indicating a systemic pattern rather than isolated missteps. The issue was neither the technology nor participants’ individual skills, but how teams were interacting with AI.

This raised two central questions. First, how can we help teams become aware of the challenges related to integrating AI into team settings? And what practical guidance do they need to bring AI fully into their discussions?

Session Two: Building Team-AI Chemistry

To help teams avoid the pitfalls they encountered in Session One we developed a three‑element framework:

1. Engage with AI as a team

2. Leverage AI’s role fluidity

3. Maintain collective ownership

To raise awareness of the importance of these elements we asked each team to go back to the Session One chat transcripts and reflect on how they had worked with the AI. To support this reflection, we introduced a practical checklist of questions to guide their analysis and help them recognize the interaction patterns limiting their effectiveness.

Here are some samples of the self-evaluation questions:

• Did we introduce ourselves as a team and explain our respective roles and expertise?

• Did we assign more than one role to AI?

• Did we articulate our reasoning clearly, or did we fall back on short, minimal responses?

By working through this set of questions, participants reflected on how they interacted with AI as a team and identified the areas where they could strengthen the interaction in the next workshop. We then provided practical tips and ready-to-use prompts designed to help teams integrate AI more intentionally as an active participant in their conversations, not merely a tool they query. For example: “You act as [role] and guide us, one question at a time to [objective]. Wait for our answer before proceeding.”

This proved helpful. After the second session, teams looked more upbeat. A detailed review of the chat logs revealed that most teams now introduced themselves in turn as a group, and the AI began factoring in the nuances of different roles and expertise, rather than treating the group as a single individual. Teams started having a collective conversation with AI.

Teams also began using AI much more flexibly, moving beyond standard fixed roles like “note-taker,” “expert,” or “analyst.” Depending on the stage of the discussion, AI was asked to act as a brainstorming partner, a challenger to test assumptions, a “prototyper” to create artifacts, and a storyteller to refine pitches. Teams realized that AI can instantly shift across stakeholders and viewpoints becoming a multi-role team member within the same meeting.

While the first two capabilities—interacting as a team and assigning multiple roles to AI—were absorbed relatively quickly, the third one, collective ownership, took longer to mature. Still in the second session, some teams had the tendency to follow the AI rather than steer it, sitting around the screen and passively reacting to its outputs. The AI seemed to lead; the team responded. By later sessions however, the dynamic had gradually shifted. Before submitting a prompt, teams paused to discuss among themselves how to frame the next iteration. They debated alternative directions, conducted judgment checks, and collectively challenged the AI’s outputs before engaging again. These pauses proved instrumental: they prevented the team from slipping into “spectator mode” and helped them remain firmly in the driver’s seat. Survey data of later sessions confirmed the benefits. Average engagement increased by 30 percent, and participants reported that the AI was providing more meaningful support to their team discussions. Two-thirds noted that their group conversations, alignment and collaboration had improved as a result, leading to higher-quality outputs. Three out of five participants noted that collective judgment mitigated typical pitfalls of using AI alone, such as too much trust or conformity.

Building Team-AI Chemistry in Your Meetings

As our experiment shows, this kind of team-AI chemistry does not come naturally, and it rarely emerges on the first attempt. For most teams, it needs to be deliberately nurtured and intentionally embedded in the way they work. The risk is that the effect fades if it is not reinforced. How to do it in practice? Here are three tips:

• Plan the meeting agenda with explicit AI slots. Identify the agenda points where AI participates and specify the role it should play: for example, plan a five-minute introduction round where the team briefs AI with context, or a 15-minute “challenge slot” towards the end of the meeting where AI plays the skeptic.

• Prepare a few prompts to summon AI in a designated role. For instance, “take the perspective of …” and “how might stakeholder XYZ react to …” or include cues that ensure judgment pauses, such as: “Wait for our decision before proceeding.”

• After the session, review the chat transcript. Check it against a checklist of questions to see how well the team-AI dynamics unfolded and identify opportunities to improve next time. You can also use AI as a coach (by uploading the chat transcript and asking for AI’s evaluation against the checklist) or sparring partner in a dialogue (by discussing ways to strengthen team-AI interaction in future sessions).

Putting these tips into practice requires more than good intentions: teams often lack the authority to redesign meetings or change established rituals on their own. Leaders play a critical role as they are the ones that can decide and embed a new way of working. A practical first step is for leaders to involve their team in an experiment, taking one team meeting and intentionally integrate AI in it. When doing this, leaders should set expectations that it may take a few iterations to climb the learning curve and master the new approach. Once the new practice is tested, leaders should continue to apply it and avoid their teams reverting to old habits.

. . .

Team-AI chemistry doesn’t develop automatically. To develop it, high-performing teams engage AI collectively, draw on its multiple roles, and maintain shared ownership of the interaction. When nurtured deliberately, these capabilities enhance team performance by yielding better alignment and coordination and ultimately elevating the quality of teamwork outcomes.

Copyright 2026 Harvard Business School Publishing Corporation. Distributed by The New York Times Syndicate.

Explore AAPL Membership benefits.