A lot, because nothing ever happens.

The iron law of "nothing ever happens"

There are decades where nothing ever happens, and there are weeks where we are so back

It seems like MS could really fuck them up if they stopped using OpenAI for all their azure stuff. As of now I don't think MS relies on their own LLM for anything?

Inshallah I am fed up of dealing with these charlatans at work

A solution in search of a problem

My guess for the dynamics: openAI investors panic, force the company to cut costs and increase pricing, other AI company investors panic, same result, AI becomes prohibitively expensive for a lot of use cases ending the hype cycle.

I've hesitated to ever write anything about it thinking it'd come across as too or Luddite, but this comment kind of inspired me to flesh out something that's been simmering in the back of my head ever since LLMs became latest fad after the NFT boom.

One of the most unnerving things to me about "AI" in the common understanding is that its entire hype cycle and main use cases are all tacit admissions that all of the professional and academic uses of it are proof that their pre-"AI" standards were perfunctory hoop jumping bullshit to join the professional managerial class, and their "artistic" uses are almost entirely utilized by people with zero artistic sensibilities or weirdo porno sickos. All of it belies a deep cynicism about the status quo where what could have been heartfelt but clumsy writing by young students or the athlete in your example are being unknowingly robbed of their agency and the humanizing future of looking back on clunky immature writing as a personal marker of growth. They're just hoops to jump through to get whatever degree or accolade you're seeking, with whatever personal growth that those achievements originally meant stripped of anything other than "achieving them is good because it advances your career and earning potential." Techbros' most fawning and optimistic pitches of "AI" and "The Singularity" instead read to me as the grimmest and most alienating version of neoliberal "end of history" horseshit where even art and language themselves are reduced to SEO marketized min/maxxed rat races.

I hope this doesn't sound too but I had to get that rant out

Maybe I'll expand that into something

I don't think it's purpose and mechanisms being non-artistic is incompatible with people finding meaning in it. We find meaning in random stuff all the time, it's kind of just our thing

I mean this is exactly part of the reason they’re going bankrupt which is good so you should keep doing it. Companies have been using other forms of AI with some success whereas LLM just regurgitates too much random fake information for anyone serious to use professionally.

If it goes under, use open source LLMs which have been steadily improving and almost surpassing proprietary ones.

Yeah but integrating LLMs with other systems is already happening.

Most recent case is out of Deepmind, where they managed to get silver medalist score in the International Mathematics Olympiad (IMO) using a LLM with a formal verification language (LEAN) and then using synthetic data and reinforcement learning. Although I think they had to manually formalize the problem before feeding it to the algorithm, and also it took several days to solve the problems (except for one that took minutes), so there's still a lot of space for improvement.