What will the Chinese Communist Party do with a Large Language Model?
1) 90% - China might even be doing it already
2) 25% - I think that too many other groups are going to put out LLMs that might not be top notch but are close
I definitely agree that we’re going to move to the right on that graph with more economic utility but I wonder if we’ll really go up. On the one hand more data + more compute sounds like it’ll always be better. But will there really not be more open source datasets from Eleuther or groups like that? I saw that ImportAI this morning said something about the cathedral versus the Bazaar where the Cathedral was OpenAI / DeepMind / GoogleBrain and the bazaar was Eleuther and other open source groups. I’d bet good money that those groups are going to stay near state of the art. But then again even if you have such a huge model how do you deploy it. Someone got LLaMa up and running on a RaspberryPi but it took 10 seconds per token.
China and India are facing a gender crisis. There is a surplus of men that are doomed to be single. I could see the CCP and the Indian government pushing lovebots onto these lonely men. After all, large groups of single men have rarely ended up well for governments.
The West is also facing a similar crisis. More women graduate from college than men, and they don’t want men without an education. A lot of non-college-educated men will end up single and the government may placate them with lovebots. It’s a disturbing thought but may one day be a reality.
Liam, under what circumstances would you consider being interviewed on or contributing this article for an online HR magazine I am editing?
if interested email me firstname.lastname@example.org