Comment by nfw2

I think the issue with legibility comes down to the fact that most users are not using LLMs for tasks where improvements to raw reasoning abilities wouldn't help much or at all. So it's not a matter of anyone's deficiency of perception but rather a lack of any benchmark to perceive.

It's kind of like how the consumer laptop market is now. I was telling my boss today that most employees wouldn't see any noticeable performance difference between a macbook pro and a neo if they are just doing admin stuff on the web.