[–] pj_mukh 1 day ago
[–] harimau777 1 day ago
The bottleneck for housing is construction and landlords. The bottleneck for medicine is the number of doctors and the inelasticity of demand. The bottleneck for education is the number of teachers. Etc.
[–] pj_mukh 1 day ago
This is patently false on multiple levels. One, on a simple level automating paperwork frees up labor for more useful things they can do immediately. Two, the footprint of what AI can do is rapidly growing. Till 2 months ago, no one thought LLM's could command robots, and now voila [1]. Of course these capabilities will plateau somewhere, but it's clear the productivity curve hasn't even been explored yet.
At least in my city, housing is severely restricted by the cost and availability of labor. I want labor to be paid well, but I always want housing to be cheaper, the missing link is productivity.
[1]: https://x.com/ihorbeaver/status/2104646736652447854?s=20
It's easy to find egregious corner cases of philosophical workflows where an LLM was inappropriately used to replace thinking, where the whole goal was thinking (which is the case here).
But we know most token use is software development + back office paperwork. A big chunk of it is productivity improvements for things we all need (housing, medical paperwork, insurance etc.), so to me, that's where my argument lies.