The World Cup final presents the last question. Spain and Argentina have reached the championship match, with the ultimate showdown on the pitch about to begin. Simultaneously, the AI prediction contest that has run throughout this tournament is entering its final stage. Twelve major AI systems have provided clearly divergent answers regarding the champion: seven favor Argentina, four support Spain, while Kimi predicts a 1-1 scoreline.
The program "Human vs. Machine: Who is the World Cup Prophet," a collaboration between Lenovo Group and Migu Video, has placed AI in the ongoing "exam" of the World Cup—a scenario with public outcomes, high process uncertainty, and mass participation. From predicting single-match results and scores to forecasting the entire "Road to the Championship" through the knockout stages, AI must not only answer "who wins this match" but also face longer chains of continuous deduction and real-world verification against actual results.
According to the championship path predictions, only three large models remain with a chance of being fully correct: Tianxi, which supports Spain, and Zhipu and SenseTime's Little Raccoon, which both back Argentina.
Final Predictions from 12 AIs Before the Final
The predictions released before the final show Argentina has garnered more AI support. Seven models—DeepSeek, Tongyi Qianwen, Baidu's ERNIE, Tencent Hunyuan, Zhipu, Jieyue, and SenseTime's Little Raccoon—favor an Argentine victory, with score predictions heavily concentrated on a 2-1 win for Argentina over Spain.
On the other side, four models—China Mobile Jiutian, Tianxi AI, MiniMax, and iFlytek Spark—choose Spain, with their representative scores also largely centered on 2-1. Kimi predicts a 1-1 draw, a result that does not directly declare a winner.
Therefore, rather than simply summarizing the 12 AIs' answers as "collectively favoring one side," a more accurate description is: Argentina holds a 7-to-4 majority, but this does not constitute an overwhelming consensus. Furthermore, the mainstream predicted score on both sides is 2-1, indicating the models' shared judgment for this final leans more towards a "clash of titans decided by a narrow margin" rather than a one-sided contest.
The Championship Path: Extending a Single Prediction into an Entire Knockout Journey
In the "Road to the Championship" segment of the human-machine contest, the prediction task extends beyond a single match. Starting from the knockout stage, AIs must predict the advancing team in each round, ultimately naming the champion. Each round's outcome influences the next round's matchups; a single error in judgment can alter the entire subsequent path. Compared to guessing the outcome of a single game, this format more closely resembles a long-chain stress test. It examines not whether an AI can occasionally guess a correct score, but whether it can string together factors like team strength, tournament structure, potential opponents, and knockout-stage uncertainty into a relatively coherent chain of judgment.
Looking back at the actual "Road to the Championship" for the two finalists, Spain defeated Austria, Portugal, Belgium, and France in the knockout rounds. Argentina eliminated Cape Verde, Egypt, Switzerland, and England. Having progressed to the final, all previous predictions about their advancement paths have been verified round by round, converging into the final question: Spain or Argentina, who can take the last step?
This makes the final predictions from the 12 AIs even more intriguing. Viewed in isolation for the last match, it's 7 vs. 4 plus one draw prediction. Placed back within the complete chain of the "Road to the Championship," one can further review: which AIs correctly identified the final matchup earlier, which exhibited more stability in predicting the knockout path, and whose champion prediction can hold firm until the end. The championship result is merely the final answer; the complete predictive path offers a more comprehensive review of predictive capability.
The Final World Cup Match: The Last Public Test for AI Predictions
The greatest charm of football lies precisely in the fact that it cannot be completely reduced to a deterministic answer by data. Squad form, in-game tactics, random events, even a single shot can alter the course of a match. AI can enhance the systematic nature of judgment based on data and historical information, but it cannot eliminate the inherent uncertainty of the game itself.
Precisely because of this, the significance of the World Cup prediction contest extends beyond "who guessed most accurately." It places different AI models into the same real-world scenario—one familiar to the public, with verifiable outcomes and continuous change—subjecting their judgment, stability, and explanatory power to public examination through each match. From this perspective, the World Cup is not a showcase for AI, but more like an open examination hall.
After the final whistle, the World Cup champion will be officially revealed, and the final answers from the 12 AIs will face their last comparison. Will Argentina, supported by seven AIs, continue its championship glory? Or will Spain, backed by four AIs, ascend to the top of the world? And could the 1-1 draw predicted by Kimi become another possibility in this ultimate showdown? The answer lies in the final 90 minutes, and any potential extra time and penalty shootout.
Comments