Comment by trashb

It is entirely plausible that your colleague did have a significant speedup from using AI. But that may also be a one off overstated story (probably not as you respect his judgement as "senior dev").

I'm not saying they didn't get a speedup or you can't get a speedup. I'm just wondering how they determined that speedup and how reliable they are as a source for the statement "using AI will improve your performance".

As the other thread was discussing, it is a difficult thing to prove but some (debatable) measurement proves more then anecdotal stories. And I am trying to probe for some sources where these kind of statements are proven as I am a programmer continually judging the tools available to me.

And on the flip side there is at least some evidence that programmers often overstate/overestimate their speedup when using AI.