Prior to the start of the 2026 FIFA World Cup, the team at Unanimous AI conducted the world’s first conversational debate among human participants and AI agents. Our objective was to connected humans and AI agents so effectively we could fosterer the emergence of a “collective superintelligence” that could accurately predict the (a) the final four teams, (b) the most likely “dark-horse” team to emerge during the tournament, and of course, (c) the ultimate winner of the cup.

As we published in our June 12th press release, we produced this set of predictions this by enabling a large group of 44 human sports fans and 11 AI agents to participate together in a real-time conversational debate in which all 55 participants (human and AI) were empowered to argue their positions with evidence and reasoning.  This was conducted using our Thinkscape platform which connects large groups into conversational systems modeled on biological swarms.  The result is a new type of “Hybrid Collective Intelligence” that efficiently combines the knowledge, wisdom and intuition of human forecasters with the depth and rigor of AI agents.

So, which teams did the Hybrid Collective Intelligence favor? 

The full conversation lasted 40 minutes.  During that time, the large group of humans and AI agents discussed each of the issues above, arguing points and counterpoints while their perspectives, evidence, and reasoning were tracked by the underlying AI engine.  At the conclusion of the discussion, the top ten candidates to win the World Cup was instantly produced:

As shown above, the TOP FOUR TEAMS were determined to be:  Spain, France, Argentina, England

As also shown , the MOST LIKELY WINNER was determined to be:  Spain

And finally, the likely DARK HORSE to stand out during the tournament: Norway

Reasoning: 

The group’s verdict is that Spain is most likely to win the World Cup this year. The group cited Spain’s exceptional squad depth, midfield control anchored by Rodri and Pedri, and Lamine Yamal as a decisive X-factor. The Hybrid Collective Intelligence also expressed that Spain’s tactical cohesion and ability to manage the tournament’s heat conditions are differentiating advantages.

France came in second in the debate with a Scaled Ranking of 2.22, followed by Argentina at 6.08. This gap is particularly striking when compared to the leading prediction markets such as Polymarket and Kalshi, where Spain and France have nearly equal odds. The collective appears to be expressing a level of conviction about Spain’s dominance that the broader market has not yet fully priced in. The group’s top four projection also includes Argentina and England, with Portugal and Brazil receiving meaningful consideration.

One of the session’s most striking findings was the group’s confidence in Norway as the lower-ranked team most likely to win matches in the knockout rounds. The group credited Norway’s elite attacking firepower, led by Erling Haaland and Martin Ødegaard, high-scoring qualifying form, and a roster of players competing at top European clubs. Concerns about over-reliance on Haaland and limited World Cup experience were debated, but the collective concluded that Norway’s individual brilliance gives them a realistic path to knockout wins.

Should we trust this forecast?

All sports predictions have a large margin of error due to the impact of chance. That said, a recently published study compared Hyperchat AI™ technology to large-scale prediction markets and found that groups of only 25 people engaged in Hyperchat AI conversations outperformed Polymarket in sports forecasts that leveraged hundreds of traders.

How does the Technology work?

The Thinkscape platform enables groups of any size to hold thoughtful conversations (by text, voice, or video) in which they discuss complex issues in real-time, freely offering arguments and counterarguments until they converge on a unified answer. Because these discussions are deliberative, with every participant expressing not just opinions but evidence and reasoning, solutions emerge that amplify the collective intelligence of the group (see research paper: https://arxiv.org/abs/2401.15109).

What was novel about these World Cup predictions is that the debate participants didn’t just include humans but also included Unanimous AI’s proprietary “Contributor Agents” that are designed to offer their own points and counterpoints within the real-time group discussion. The goal was to enable teams of humans and AI agents to discuss issues so collaboratively that they form a hybrid collective intelligence that combines human wisdom with AI rigor. And these contributor agents were highly valued by the human participants, with over 90% of users saying they were helpful to the group deliberation.

“What you’re seeing here is the first time a large group of humans and AI agents have ever held a real-time meeting together,” said Dr. Louis Rosenberg, CEO and Chief Scientist of Unanimous AI. “When you bring together 44 sports fans who intuitively know football and have them mix-it-up with each other and with AI agents who view the sport with statistical rigor, the output isn’t just a list of opinions — it surfaces the most compelling evidence and reasoning — a true hybrid intelligence of humans and AI.”

A short video demonstrating these results is available at https://youtu.be/YpLm7moMD_Y

Learn more about how Thinkscape makes teams smarter visit Thinkscape.ai.