@ElliotGlazer Is it more likely they will update the Lean formalization with the natural language version they are claiming to solve? Or update the papers to reflect the theorems proved in Lean?
@GregHBurnham Just to clarify, do you think if the benchmark is 100% saturated, AI can be a drop in replacement? Or are there tasks not fully captured in the benchmark
@fchollet Do you see this as a limitation with RSI looming? Wouldn't the objective be that RSI would be able to discover much more efficient algorithms that this would not be an issue?