A British AI’s top-ten finish in a global forecasting competition is accelerating a shift in thinking within the industry: the future isn’t about humans versus machines, but humans with machines. ManticAI’s eighth-place performance in the Metaculus Cup demonstrated the power of AI, but experts believe the ultimate key to prediction is collaboration.
The contest pitted entrants against 60 unpredictable real-world events, from political leadership races to the path of environmental disasters. While ManticAI outperformed many human professionals, the very top spots were still claimed by elite human “superforecasters,” highlighting the distinct strengths that both sides bring to the table.
Warren Hatch, the CEO of Good Judgment, a company co-founded by forecasting expert Philip Tetlock, summarized the sentiment perfectly. “The main point for us is that the answer isn’t human or AI, but instead human and AI,” he said. He envisions a future where AI excels at data-heavy tasks, like predicting monthly inflation rates, while humans handle questions that require more judgment due to sparse data.
ManticAI’s system itself is a model of collaboration, using a team of different AI agents to analyze problems. This multi-agent approach, combined with its ability to work persistently, allows it to cover more ground and react to new information faster than any single human could. However, experts note it can sometimes lack the ability to spot logical inconsistencies in its own output, a skill where humans still excel.
This vision of a hybrid future was echoed by Lubos Saloky, a human forecaster who came in third. “I do not plan to retire,” he stated. “If you can’t beat them, merge with them.” This practical approach signals a move towards a new standard in forecasting, where human intuition guides and verifies the powerful analytical engines of AI to achieve unprecedented accuracy.
