I work in AI, and the fatigue is real.

What I've found most painful is how people with no fucking clue about AI or ML chime in with their expert advice, when in reality they're as much an expert on AI as a calculator salesman is an expert in linear algebra. Having worked closely with scientists that hold PhD's, publish papers regularly, and who work on experiments for years, it makes me hate the hustle culture that's built up around AI. It's mostly crypto cunts looking for their next scheme, or businesses looking to abuse buzzwords to make themselves sound smart.

Purely my two-cents, but while LLM's have surprised a lot of people with their high quality output. With that being said, they are known to heavily hallucinate, cost fuckloads, and there is a growing group of people that wonder whether the great advances we've seen are either due to a lot of hand-holding, or the use of a LOT of PII or stolen data. I don't think we'll see an improvement from what we've already seen, just many other companies having their own similar AI tools that help a little with very well-defined menial tasks.

I think the hype will die out eventually, and companies that decided to bin actual workers in favour of AI will likely not be around 12-24 months later. Hopefully most people and businesses will see through the bullshit, and see that the CEO of a small ad agency that has positioned himself as an AI expert is actually a lying simpleton.

As for it being "real AI" or "real ML", who gives a fuck. If researchers are happy with the definition, who are we to be pedantic? Besides, there are a lot of systems behind the scenes running compositional models, handing entity resolution, or building metrics for success/failure criteria to feed back into improving models.