The story of AI in forecasting is one of astonishingly rapid improvement, captured by a single statistic: in just one year, the top-performing bot has leaped from a rank of around 300th to 8th place in a major international competition. This incredible surge, led by British startup ManticAI, is a stark indicator of the exponential progress in artificial intelligence.
This dramatic ascent was a key takeaway from the latest Metaculus Cup. Professional forecaster Ben Shindel, who witnessed this evolution firsthand, highlighted the difference. “We’ve really come a long way here compared with a year ago,” he said, reflecting the community’s amazement at the pace of change.
This isn’t a case of slow, incremental gains. It is a disruptive leap in capability, driven by advances in large language models and the sophisticated ways companies like ManticAI are orchestrating them. By using a team of AI agents to tackle problems, they have unlocked a new level of analytical power.
The implications of this performance surge are profound. It suggests that the timeline for AI reaching and potentially surpassing human-level performance in complex cognitive tasks may be shorter than many have anticipated. What seemed like a distant possibility is now becoming a near-term reality.
While the AI still has areas to improve—particularly in complex logical reasoning—its current trajectory is undeniable. This leap from the middle of the pack to a top contender in just twelve months has put the entire field of forecasting on notice: the machines are not just coming; they are arriving with breathtaking speed.