There wasn't ever a ChatGPT or Perplexity.ai moment for most technical tasks. I'm always evaluating these models to make sure that the people I interview remot

There wasn't ever a ChatGPT or Perplexity.ai moment for most technical tasks. I'm always evaluating these models to make sure that the people I interview remotely for my company don't cheat. There are certain questions the models are good with, but they will often produce very predictable answers (a lot of lists for open-ended questions). On the other hand, for many specific technical questions, they will predictably give wrong answers. Perplexity.ai will fail to find some of the same content that Google would if they indexed the same blog or documentation. Both ChatGPT and Perplexity are horrendous at reasoning within day-to-day knowledge of any senior knowledge worker, there just aren't enough parameters. Therefore, they often can't help with questions in that context.