I am not sure if I read the correct thread, but I personally didn’t find your arguements convincing, although I think a full ban is excessive (at least initially).

Keep in mind that I do use local LLM (as an elaborate spell-checker) and I am a regular user of ML based video upscaleling (I am a fan of niche 80s/90s b-movies).

Forget the technical arguments for a seconds. And look at the social-economic component behind US-style VC groups, AI companies, and US technology companies in general (other companies are a separate discussion).

It is not unreasonable to believe that the people involved (especially the leadership) in the abovementioned organizations are deeply corrupt and largely incapable of honesty or even humanity [1]. It is a controversial take (by US standards) but not without precedent in the global context. In many countries, if you try and argue that some local oligarch is acting in good faith, people will assume you are trying (and failing) to practise a standup comedy routine.

If you do hold a critical attitude and don’t buy into tedious PR about “changing the world”, it is reasonable to assume that irrespective of the validity of “AI safety” as a technical concept, the actors involved would lie about it. And even the concept was valid, it is likely they would leverage it for PR while ignoring any actual academic concepts behind “AI safety” (if they do exist).

One could even argue that your arguementation approach is an example of provincialism, group-think and generally bad faith.

I am not saying you have to agree with me, I am more trying to show a different perspective.

[1] I can provide some of my favourite examples if you like, I don’t want to make this reply any longer.