Microsoft funds AI research on team collaboration

Microsoft Research is funding Spring 2026 AI studies on how teams work better together.

Microsoft Research has opened a Spring 2026 call for proposals focused on one specific problem: AI helps individuals a lot more easily than it helps groups. The program is aimed at university researchers, with awards of about $50,000 to $75,000 and a fast review cycle.

Microsoft is betting on group AI, not solo AI

The core idea behind the AI and the New Future of Work CFP is refreshingly specific. Microsoft Research wants work that shows how AI can help a team outperform a person using AI, or a team without AI, on real tasks.

That focus matters because a lot of current AI progress is measured at the individual level. A single person can draft text, summarize meetings, or generate code faster with tools like OpenAI products, Anthropic models, or Microsoft’s own Copilot. But a team is a different problem: multiple people, shifting context, uneven expertise, and plenty of room for noise.

Microsoft’s CFP is built around that gap. The company is asking researchers to study how AI changes collaboration itself, not just individual productivity. That includes who gets heard, when an assistant should speak up, and how to reduce the low-value work that clogs group projects.

• Systems that help a team with AI outperform an individual with AI
• Simulation environments and world models for collaborative settings
• AI systems that cut collaboration drudgery and attention overload
• New norms for AI-embedded teams and better common ground maintenance

The proposal topics read like a research agenda for real offices

The topic list is broader than a standard productivity grant. Microsoft is asking for work on collaboration arenas, behavioral science, AI proactivity in group conversations, and ways to make onboarding and stepping away from projects less painful.

There is also a strong social angle. One section asks for non-exploitative AI systems where the value of data created by a team flows back to the people who created it. That is a more interesting question than the usual “how do we automate this task” framing, because it gets into incentives, ownership, and trust.

In plain terms, Microsoft is asking researchers to study the messy parts of teamwork that software usually ignores: interruptions, handoffs, low-value meetings, overloaded chat channels, and the awkward moments when an AI should speak up or stay quiet.

“The most important thing is not to automate the task, but to redesign the work,” said Eric Horvitz, Microsoft’s Chief Scientific Officer, in a Microsoft Research discussion about AI and work.

That quote fits the spirit of this CFP. Microsoft is not asking for another assistant that writes emails a little faster. It is asking for evidence that AI can change how groups coordinate, decide, and share attention.

The money is modest, but the bar is high

The funding size tells you a lot about the kind of research Microsoft wants. At roughly $50,000 to $75,000 per award, this is enough to push a focused academic project forward, but not enough to support a large lab program by itself.

Applicants are also being asked to keep proposals short: one page, 500 words plus references, with budget details included. Microsoft says the review cycle will be quick, and it wants results that can make material impact quickly.

• Eligible applicants must come from academic institutions or non-profit research organizations
• Proposals must fit university policies for unrestricted gifts
• Incomplete proposals or requests above the maximum award will be excluded
• Microsoft says it may use names and likenesses to publicize selected proposals

That structure makes the CFP feel less like a slow academic grant and more like a targeted experiment. It rewards teams that already have a concrete research question, a method, and a plausible path to results within a short timeline.

It also hints at Microsoft’s practical interest. The company is not funding speculative essays about AI and society in the abstract. It wants studies that can produce evidence about how AI should behave in meetings, group chats, project handoffs, and decision-making sessions.

This CFP is a signal about where workplace AI is headed

If you read this announcement carefully, the real story is not the budget or the dates. It is the shift from individual copilots to collective intelligence tools that understand teams as social systems.

That is a harder problem, and probably a more valuable one. A model that helps one person write faster is useful. A model that helps five people avoid duplicated work, missed context, and meeting fatigue could change how organizations operate.

The next useful benchmark for workplace AI may not be “how well does it answer a prompt?” It may be “does the team finish better work with less friction?” If Microsoft’s CFP attracts strong proposals, expect more research trying to answer that question with real data instead of product demos.

For researchers, the actionable move is simple: if your lab studies collaboration, negotiation, organizational behavior, or human-computer interaction, this CFP is worth a close read before the May 25 deadline. For everyone else, it is a good sign that the next wave of AI tools may care less about solo output and more about how people actually work together.