[d136o]

I’m far from an expert (just an engineer who likes to download and try out demos) but one dark view of the alignment chatter is that it’s really about whether can you can get this new big artificial brain to push any user in the direction of BigCorp goal, while letting the users think that it’s answering their questions etc.

I’d love to think that we can have AI agents working locally (with privacy) in alignment with our personal goals, but that seems like one of those things that won’t happen. (Will apple come through in terms of enabling local+decentralized inference/training?)

[ralfn]

It's in the interest of the semi conductor industry for LLM to become a commodity, so it will.