A common criticism of artificial intelligence is that it can only find patterns in old data, not generate truly original ideas. However, an AI forecaster’s recent success in a global competition is challenging that notion, suggesting that machines are now capable of making predictions that are, in a sense, “original.”
The AI, from UK startup ManticAI, placed eighth in the Metaculus Cup. According to its co-founder, Toby Shevlane, its performance was noteworthy because it didn’t just follow the human pack. “You could say our system’s predictions were more original than most human entrants,” he said, “because people often cluster around the community average predictions. The AI system often strongly disagreed.”
This originality stems from the AI’s freedom from human cognitive biases. It does not feel the social pressure to conform to a consensus. Its conclusions are the product of its own systematic analysis, which involves using a team of AI agents to research, model scenarios, and synthesize findings from the ground up.
This ability to generate a well-reasoned, independent, and often contrarian viewpoint is a form of originality. It is not creating something from nothing, but it is creating a novel synthesis of information that leads to a non-obvious conclusion. And as its high ranking demonstrates, these original predictions are often correct.
This capability has profound implications. It means AI can be used not just to automate known processes, but to discover new insights and challenge long-held assumptions. Whether in science, business, or policy, an AI that can produce an accurate, original forecast is an incredibly valuable tool for innovation and progress.