“One million high-paying upper-middle class family supporting jobs were destroyed, but on the bright side, one million minimum-wage Walmart greeter jobs were created! See? We’re breaking even!”
Building and maintaining data centers aren’t minimum wage jobs. They are the best kind of jobs: accessible to somewhat above average people across a broad range of backgrounds. (As opposed to upper middle class knowledge worker jobs that are only accessible to the cognitive top 10%.)
I understand building data centers, but what does maintaining data centers entail? Are those going to be high paying jobs and how many of them. I don’t know much about the area but it seems like they would be low wage and not very many needed once the data centers are built.
I think we are kidding ourselves if we think writing software was only accessible to the cognitive top 10%. And even if that were the case for the highly paid software jobs, there was a long tail of WordPress customization type jobs that required less experience and will probably be going away.
you seem to be discounting or unaware of the possibility of a) automating a significant chunk of new buildouts, and b) domestic companies seeking a slice of the “sell open model hosting for less than competitors” pie.
I truly believe the cost per token for most people will eventually go approximately to zero even in the absence of subsidy unless there is a dramatic state intervention to curtail free computing and hardware R&D and manufacturing.
Fair callout. Strongly discounting (b) and (a) is underspecified for me to respond reasonably.
Your premise of full token commoditization has a major wrinkle though. GPU-centric data centers with never allow tokens to go to zero since you have a 1.5 year technical deprecation and 5 year life on the GPUs. Maybe Cerebras or similar chip builders, after writing model weights to silicon, will improve those returns on capital and enabling the scaling you are envisioning.
From my perch, AI at the edge is severely underserved, and the GTM for the hyperscalers mostly ignore it. So if someone figures out how to scale AI at the edge (nvidia + HF perhaps) then I think your premise is on point.
There is this overwhelming desire on the part of many people to ascribe agency to LLMs when what the LLM really is is simply not a magic world model but a stochastic text generator. For some reason that truth haunts them and they have a deep compulsion to prove it wrong for feelings reasons. I can only surmise why this afflicts some people and not others and I’m sure that one day there will be fascinating psychological studies explaining such phenomena.
In the future (ha ha ha) when everyone has become totally LLM dependent a service outage will cause all of them to stop walking and talking and otherwise functioning and their heads will simply droop down while they stand or sit silently in place.
reply