AI EXPERTS DRAMATICALLY UNDERESTIMATED FIELD'S PACE

The study reveals significant gaps between expert predictions and actual AI progress. AI reached gold-medal performance at the International Mathematical Olympiad five years before the median expert forecast. Anthropic's annualized revenue reached approximately five times what forecasters predicted. However, real-world applications show a more uneven trajectory. Predictions for practical deployments like autonomous vehicles remain less accurate, suggesting experts struggle more with timeline estimates for applied AI versus fundamental capabilities. The findings highlight the difficulty of forecasting technological progress in rapidly evolving fields. As AI development accelerates, the gap between expert consensus and actual advancement continues to widen, raising questions about how well current predictions for future AI capabilities reflect realistic timelines.