I always get a kick out of walking past a crowded bar and watching avid fans go crazy as their favorite NCAA basketball team notches another win.
The confusing New York springtime weather (sometimes it’s frigid, sometimes it’s weirdly humid) doesn’t deter my fellow inner-city sports fanatics from supporting their teams in an electrifying public forum. As the 2026 March Madness tournament begins and everyone starts filling out their brackets, everyone turns into the biggest college basketball experts as they zero in on who’s going to emerge as this year’s champions.
Article continues below
The four No. 1 seeds are the favorites to win: Duke, Michigan, Arizona and Florida. But only one can prevail — let’s see if these chatbots’ picks line up with the bet takers out there.
ChatGPT
With the prompt: Which team do you think will win the 2026 Men’s NCAA Tournament and why?), I kicked off my March Madness 2026 prediction experiment with ChatGPT. It ultimately chose Duke and highlighted four of the main reasons why it went with that team:
- They’re the consensus No. 1 team: Duke entered the tournament as the No. 1 overall seed and finished the regular season ranked No. 1 in the AP poll
- The odds back it up: Sportsbooks have Duke as the favorite (+330) — slightly ahead of teams like Michigan and Arizona
- Elite balance (the real championship indicator): Duke is consistently described as having top-tier offense + defense, which historically correlates strongly with title winners
- Path + depth: Depth matters in March — Duke tends to have multiple NBA-level contributors, reducing upset risk
Even though it predicted Duke as the ultimate victors, ChatGPT also pointed out four other teams that could overcome Duke and shock us all. Those teams included Michigan, Arizona, Florida, Houston, and Gonzaga. Finally, ChatGPT foresaw Duke defeating Arizona in the title game to take it all.
Gemini
Next up is Gemini’s turn to select the 2026 Men’s NCAA Tournament victors. Once again, Duke came up as the projected winner. Gemini’s reasoning for its pick spotlighted Duke’s star player, their offensive strengths, and their top-tier coaching:
- The boozer factor: Freshman Cameron Boozer is playing at an All-American level, averaging a double-double (22.5 PPG, 10.2 RPG). His ability to stretch the floor and protect the rim makes Duke nearly impossible to match up with
- Offensive firepower: They rank in the top 5 for adjusted offensive efficiency. In a tournament where games often grind to a halt, Duke has multiple “bucket-getters” who can create their own shots
- Coaching maturity: Jon Scheyer has finally found the perfect balance between high-end “one-and-done” talent and reliable veteran leadership
Gemini also mentioned that the biggest opposition to Duke’s championship win is Purdue since they beat Michigan to win the Big Ten Tournament and have momentum on their side. Plus Gemini propped up Purdue’s guard Braden Smith as a force that is recognized as “arguably the best floor general in the country.”
Perplexity
Next, I prompted Perplexity. This AI assistant surprised me as it went in a totally different direction with its pick by not choosing Duke. It actually went for Arizona (with Duke as a very close second, which was unsurprising). Perplexity’s reasoning revolved around the results given by quantitative models, analysts, 25 coaches who were polled about their pick, and betting/prediction markets:
- Multiple quantitative models have Arizona, Duke and Michigan in a near three‑way tie, with Arizona a slight favorite overall, helped by a friendlier West-region path and travel advantages
- Analysts and anonymous coaches consistently highlight Arizona’s balance: top‑five level offensive and defensive efficiency, dominance on the glass, and real depth, which tends to matter more as legs get tired late in March
- Betting and prediction markets price Arizona among the very top tier of contenders (along with Duke and Michigan), implying a nontrivial but still modest title probability in a wide‑open field
Claude
And finally, I prompted Claude for its prediction.
To the surprise of no one, it prophesied a victory for Duke. Claude went in that direction as it stated how Duke is the most dominant team in the country right now, the fact that they only dropped two losses across a total of 34 games for the season, their superior coaching under the direction of Jon Scheyer, and their versatile lineup of elite players.
Claude left me with this final statement to highlight why Duke is going to win it all: “A 32-2 record, elite coaching, and ACC championship pedigree make them the team to beat. That said, March Madness is famously chaotic — Arizona is just as capable of winning it all, and you can never count out UConn or Gonzaga when the bracket opens up. But if I had to pick one team to put my money on, it’s the Blue Devils.”
Verdict: AI predicts Duke to become the 2026 March Madness champions
ChatGPT, Gemini, and Claude all zeroed in on Duke as the team they see coming out on top. I was shocked to see Perplexity set itself apart from its fellow chatbots by going with Arizona instead.
Seeing as how the main four projected winners are Duke, Michigan, Arizona and Florida, it came as no shock that all four chatbots predicted a winner from that group. So if you’re rooting for Duke this year, chances are you’re not cheering for the winning team.
Follow Tom’s Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds.
More from Tom’s Guide
Back to Mobile Cell Phones
