Resolved according to https://robustbench.github.io/#div_cifar10_Linf_heading. In case Robust Bench stops being active or better attacks/defenses are known which are not listed on Robust Bench, SOTA will be the resolution criteria.

The question resoves to 2030+ if it has not reached 80% for Jan 1 2031 and there is no compelling proof that this is an impossible task.

See https://bounded-regret.ghost.io/ai-forecasting-one-year-in/ for a related discussion.

I will not bet in this market.

AI Forecasting: One Year In

Last August, my research group created a forecasting contest [https://bounded-regret.ghost.io/ai-forecasting/] to predict AI progress on four benchmarks. Forecasts were asked to predict state-of-the-art performance (SOTA) on each benchmark for June 30th 2022, 2023, 2024, and 2025. It’s now past June…