Humans + AI Agents = “Collective Superintelligence”
Unanimous AI brought together 54 sports fans and 26 AI agents in a real-time conversational debate
The English Premier League kicks off this week. It’s a 10-month marathon season where every team is subject to all of the variables that professional sports brings: winning and losing streaks, injuries, questionable referee calls, player transfers, and even mid-season coaching changes. Defending League champion Arsenal are the odds-on favorite to win again, though the powerful London club hasn’t been a repeat champion of the top English league since the 1930s. So, if the 2026 club is unable to have a history-making season, who else might win?
To answer this question, we invited 80 participants (54 human sports fans and 26 AI agents) into our Thinkscape platform and asked them to have a giant conversational debate about which team was most likely to challenge Arsenal for the title this year. We know this sounds impossible – how can 54 people have a debate, let alone 54 people and 26 AI agents?
We do this using our patented Hyperchat AI engine inside Thinkscape. It enables groups of any size (dozens, hundreds, or even thousands of people) to hold a thoughtful real-time conversations and converge on solutions as a single unified collective intelligence. And with our new (Co)agent™ technology, these large groups can include AI agents that bring in helpful facts and evidence to support the human debaters.
The result is a “collective superintelligence” that has been shown in published studies to out-predict the Vegas odds, the large prediction markets, and professional handicappers. Which brings us to the big question – which team is most likely to challenge Arsenal this year for the EPL title?
The answer may be surprising to some: Chelsea.
Yes, after a large-scale discussion and debate, the group chose Arsenal’s cross-town rivals. Although Chelsea are one of the biggest clubs in the Premier League, they struggled last season to a mid-table 10th-place position, and haven’t won a title since 2017. Despite this, the group’s 19% prediction for Chelsea to win is almost double the percentage chance that the prediction market Kalshi gives for that same outcome (11%). And the group gave a noticeably smaller chance of that outcome (just 15%) to traditional EPL powerhouse Man City, which Kalshi has at 20%.
Another Question:
This group was also asked to predict the team most likely to get relegated this year. In the video below, we show how this group of 54 sports fans and 26 Agents discussed and debated the issue, offering valuable information on injuries, previous season standings, summer transfers, and coaching changes.
As you can see, the group converged on Hull. Making a complex forecast like this is incredibly difficult. In the EPL there are 20 teams and every one of them has variables on personnel, scheduling, and prior outcomes to consider. In this engagement, the AI agents were able to give human participants hard information – transfers in, transfers out, injury news, etc – that helped the group of fans to feel more informed and feel more confident in their predictions.
This ability to help human decision-makers to arrive at better-informed outcomes is a functionality built into Thinkscape. And this same technology that assists a group of sports fans can also be used for strategic business forecasts, project development or prioritization, and market research among other applications.
Want to learn more about how Thinkscape empowers large teams to deliberate at scale? Visit Thinkscape.ai.
