Can an algorithm be more original than a human? A British AI’s performance in a forecasting competition suggests the answer may be yes, precisely because it is free from the cognitive biases and social pressures that constrain human thought. This “originality of the algorithm” could be AI’s greatest contribution to decision-making.
ManticAI’s system secured eighth place in the Metaculus Cup, and its co-founder, Toby Shevlane, argued that its predictions were “more original than most human entrants.” The reason? It frequently and “strongly disagreed” with the consensus view that human participants tend to form.
This originality is a direct result of its nature. The AI does not experience confirmation bias, where we favor information that confirms our existing beliefs. It is not susceptible to anchoring bias, where we over-rely on the first piece of information we hear. And it feels no pressure to conform to the group, the bias known as groupthink.
By running its own systematic analysis based on data and models, the AI generates a forecast that is a pure reflection of its process, untainted by human psychological shortcuts. Its high ranking proves that this unbiased, “original” perspective is often a more accurate one.
This doesn’t mean the AI is creative in the human sense. But it does mean that it can produce novel and useful conclusions that a group of humans might have missed. In a world struggling with polarization and echo chambers, the ability of an algorithm to provide a truly objective and original viewpoint is an incredibly powerful asset.