British AI startup ManticAI outperforms humans in forecasting contest
ManticAI, a UK-based AI startup, ranked eighth in the Metaculus Cup, showcasing AI's growing edge in event prediction and sparking debate about its future dominance over human experts.
A British artificial intelligence startup, ManticAI, has outperformed human forecasters in the Metaculus Cup, an international competition requiring predictions for 60 future events. The AI system secured eighth place, surpassing many professional forecasters and signaling the rapid advancement of AI in forecasting.
Key Highlights
-
ManticAI, co-founded by a former Google DeepMind researcher, demonstrated the potential of AI to rival human expertise in forecasting. The competition included diverse predictions, from political outcomes (e.g., Kemi Badenoch’s Conservative leadership status) to environmental events (e.g., US wildfire acreage).
-
Ben Shindel, a professional forecaster who trailed ManticAI at one point, noted the dramatic improvement in AI forecasting. A year ago, the best AI ranked around 300th—now, it competes with top humans.
-
Deger Turan, CEO of Metaculus, acknowledged AI’s progress but emphasized humans still outperform AI in complex, interrelated predictions. Turan predicts AI could match elite human forecasters by 2029.
How ManticAI Works
-
The system uses a multi-model approach, delegating tasks to specialized AI agents (including OpenAI, Google, and DeepSeek models) for research, scenario analysis, and prediction.
-
Toby Shevlane, Mantic’s co-founder, argued AI forecasting avoids human groupthink by generating more independent predictions than consensus-driven human forecasters.
Human vs. AI Forecasting
-
Philip Tetlock, co-author of Superforecasting, published research showing humans still outperform AI on average. However, AI excels in data-heavy predictions (e.g., inflation rates).
-
Warren Hatch of Good Judgment Inc. advocates for human-AI collaboration, calling it the best path forward for accurate forecasting.
-
Lubos Saloky, the Metaculus Cup’s third-place finisher, humorously remarked, “If you can’t beat them, merge with them.”
Conclusion
While AI is closing the gap, human intuition and logic checks remain critical—for now. The competition highlights the accelerating role of AI in forecasting, with startups like ManticAI leading the charge.
Related News
TTD-DR AI Research Agent Mimics Human Iterative Writing Process
Google introduces TTD-DR, a deep research AI agent that models report writing as a diffusion process, outperforming existing systems in long-form research and multi-hop reasoning tasks.
AI agents may redefine hotel customer loyalty study finds
Florida Atlantic University researchers suggest AI agents could shift customer loyalty from hotel brands to algorithms in the hospitality industry.
About the Author
Dr. Emily Wang
AI Product Strategy Expert
Former Google AI Product Manager with 10 years of experience in AI product development and strategy formulation. Led multiple successful AI products from 0 to 1 development process, now provides product strategy consulting for AI startups while writing AI product analysis articles for various tech media outlets.