2 ms·
This is the interesting case with AI. How does a model know when a user is going too far? It really cannot. Not without reading their mind anyway. This will be
by giancarlostoro 7mo ago
This is the interesting case with AI. How does a model know when a user is going too far? It really cannot. Not without reading their mind anyway. This will be a problem for many years to come, and sadly many valid use cases will be dismissed.
This might eventually become moot once local and open source models become more common. Today's 32GB of VRAM is tomorrow's low tier gaming GPU.
- manyatoms 7mo agoIn 10 years it'll probably be affordable to have 1TB of (v)ram in a home computer. Will the model sizes keep getting bigger such that the large providers have much of an infra moat compared to local home inference?
- bakugo 7mo agoHave you been paying attention to hardware prices recently...?
- giancarlostoro 7mo agoIf eventually the home models are good enough that you dont even need a cloud AI provider, what happens to their bottom line? Local AI can do local tool calling just like Claude Code can do.