It’s very difficult for me to wrap my head around people that aren’t blown away by both this and generative image AIs, and I tend to be in the skeptical side when it comes to fancy tech adoption.

It is astonishing, even more so that some in this camp are smart people.

I chalk it up to "hype aversion" -- an internal heuristic that anything with so much hype around it must be bullshit. For some, there might also be good old resistance to change, to the unknown consequences of all these developments.

The people who are least impressed by GPT seem to be the ones who've never tried to build anything similar, even a text adventure game.

Industry outsiders also tend to take ML for granted, even when the 'experts' can see it's completely inappropriate. (My GF, a few years ago: "Why do we still need judges and juries? Can't computers judge cases and author legal opinions?" Me, at the time: "LOL, no, that would be ridiculous, and here's why..." Me, today: "Uh, about that argument we were having...")

I find generative AI very impressive, but impressive in the sense of "wow, it's impressive that an algorithm can do that." If I ignore the fact that something was algorithmically generated, then I'm not impressed by the generated output on its own.

How is #3 supposed to work when #2 applies?

Especially when folks are already out there LLM-generating SEO-spam, so even "well the answer also came up in a Google search" isn't a sign of accuracy.

Example: I want to buy some construction material that I've seen somewhere. I don't know what to search for since I don't know the term. Prior to LLM, the only way was to ask someone familiar with it, e.g. a plumber. Now I can ask GPT. And then I just do a quick image search to verify.

This doesn't always work, as some stuff is harder to verify. But what's valuable to me is it allows me to ask a vague question and have follow-ups. I usually learn just by having the conversation.