Post

increasingly believing much of the communication gap between LLM Enthusiasts and LLM Skeptics on this website boils down to “are you someone who has largely experienced a *reasonable* amount of job security in tech, or are you someone who has not experienced job security and likely aren’t in tech”

I think this is super real, and I want the conversation to move here. The longer we stay in the "is it useful" debate, the longer we defer taking action to protect ourselves in a changed world.

I am pretty concerned about a world where there's only 2-3 companies that can run these models. I have been spending the last few days idly musing about a coop that sets up hardware and runs the open models.

Bit premature maybe? If the big labs are losing money, a coop is both impractical (tokens at cost will be too expensive) and it is everybody's civic duty to help big labs lose more money by (ab)using free allowances and maxing paid plans.

I suspect that this is, maybe, true for classic chatbot inference but not for reasoning models. It comes from the logic of the race too: better incrementally spend 2x on compute than be caught with your pants down being 2% behind on the benchmarks.

Considering everything is metered by the token I can't see why that would be true? Like, if you use a reasoning model you are either using more of your quota or straight-up being charged more directly. It shouldn't change anything meaningfully.