The famous quote about the future being “unevenly distributed” perfectly describes the current state of forecasting, as a British AI demonstrated capabilities that seem straight out of science fiction, while human experts still dominate the top spots. ManticAI’s eighth-place finish in the Metaculus Cup shows that the future of AI-driven prediction is here, but it hasn’t completely taken over yet.
The competition was a microcosm of this uneven distribution. On one hand, you have ManticAI and several other bots showcasing stunning progress, leaping from a rank of 300th last year into the top tier. Their ability to relentlessly analyze dozens of problems at once represents a futuristic capability that is now a reality.
On the other hand, the podium was still occupied by humans. Elite “superforecasters” like third-place finisher Lubos Saloky continue to prove that the most refined human judgment remains the gold standard, particularly for complex, nuanced questions where data alone isn’t enough.
This creates a hybrid present, where different levels of forecasting capability coexist. For data-rich, high-frequency predictions, AI is likely already superior. For ambiguous, high-stakes geopolitical forecasts, human expertise remains paramount. The cutting edge, as practiced by firms like Good Judgment, is already moving to a model that combines both.
The performance of ManticAI is a powerful signal that the distribution is shifting rapidly. The CEO of Metaculus predicts AI parity with the best humans by 2029. So while the future isn’t fully here for everyone just yet, the results of this competition show that it’s arriving much faster than many expected.
