Benjamin Todd Says FrontierMath Hits 94%, Three Years Ahead of His Forecast
Summary
AI progress outran nearly every 18-month forecast, Benjamin Todd writes, including his own and the superforecasters'. Epoch's FrontierMath benchmark hit 94% by September 2026, more than three years ahead of his 97% by 2030 prediction. He puts median AGI around 2030, with the early 2030s as a real scenario.
Key Points
- Superforecasters expect $90bn combined annualised revenue from OpenAI and Anthropic by end of 2026. The figure reaches $100bn by July.
- Todd gives a 7% chance AI solves an unsolved maths problem by December 2027. A 2024 survey of AI researchers puts the median at 2054.
- Todd forecasts a six-hour METR 80% time horizon by end of 2026. Most expect three hours.