> People should realize we’re reaching the point where LLMs are surpassing humans in any task limited in scope enough to be a “benchmark This seems like a bold

> People should realize we’re reaching the point where LLMs are surpassing humans in any task limited in scope enough to be a “benchmark This seems like a bold statement considering we have so few benchmarks, and so many of them are poorly put together.