I am absolutely on board with the LLM inevitablism. It seems inevitable as you describe it. Everyone will use them everyday. Like smartphones.

I am absolutely not on board with AGI inevitablism. Saying “AGI is inevitable because models keep getting better” is an inductive leap that is not guaranteed.

100%. AGI is some arbitrary threshold that has nothing to do with LLMs and their broad application, rapid improvement and exploding ecosystem. The only comparable shift in memory is the advent of the Internet, and that was much slower for obvious reasons (everyone needs a computer, needs to use the phone line, needs to upgrade to faster internet, etc., etc.)

LLMs are just here and accessible to everyone with all kinds of uses. And people that hate them and don't want LLMs shoved in their face: it doesn't matter. Chatbots were already a thing for customer service, now they are almost implicit. SaaS in general is going to be completely revamped. Huge companies are going to have a shift like they did when they first digitized paper files. This is seismic and a large-scale structural upgrade to many institutions.

And that is without giving a shit about public facing LLMs or Ghibli art or writing your term papers for you. It doesn't have anything to do with AI making you dumb or being shoved into products. That is random noise, big picture. Anyone that can't see the real changes to business and knowledge retrieval isn't thinking this through or is actively burying their heads in the sand.

I doubt that LLM will keep getting better, too. Or at least, not in an economically sustainable way

Yes, people often conflate the two. ChatGPT is nice for some coding things and asking what you want for dinner. That does not mean we're all out of a job and our society has to be radically rethought.