This makes me feel that open-source LLMs that can be estimated efficiently, and run on local hardware, is an urgent need. Otherwise we will live in a cyberpunk dystopia with a centralized company that knows everything about you.
I hope (naively?) that we never have a world with actual LLM therapy.
> open-source LLMs that can be estimated efficiently, and run on local hardware, is an urgent need
It’s really quite expensive otherwise, but considering Alexa et al aren’t truly on device yet, and LLM are much more complex… it’ll be a while.
Microsoft have positioned themselves very well to own the entire stack of PC-like software development and operation.
