Quoting the following tweet:

What's your rough probability on "In 2026, it will seem as though deep learning hit a non-transient wall at some point in 2023-2025, after the relatively impressive results from GPT-4?"

In January 2027, I will compare the benchmarks of the best deep learning model of 2026 to the benchmarks of the best model of 2025. If it's apparent that no significant progress has been made in 2026, I will resolve this question to No. Otherwise I will resolve it to Yes.

The definition of "significant" will depend on what the best deep learning benchmarks of 2026 are and thus cannot be provided in advance. In case of a disagreement among market participants I will seek the opinion of well known AI experts in helping resolve this.

This market never resolves to N/A.