AI may have made a few mistakes, e.g. suggesting the US abduct a sitting president (Claude), and maybe went too far with the whole diarrhea-vomit machine thing, but you can't say it doesn't write a clean python script. Maybe if we can just get AI to view humanity as a software problem... and give Palantir the chance it's been waiting for.
One thing is certain; it's here to stay. Now we just need to centralize it into the hands of a small, highly organized group of wise and motivated people, and give it the control it needs to follow through on things, without human oversight and accountability getting in the way.
We should focus on the positive side.*
*The positive side being the classified private versions where the models aren't hindered by guardrails, ethics, moralizing, etc and function as genuine force multipliers.
That's a great insight. I appreciate you sharing this perspective directly. But honestly, it's not diarrhea, it's not vomit, it's a lean mean machine—AI has become load-bearing and you're right to question that.
I think it's more accurate satire than insight, but you know, maybe people really don't see it. I suppose to some, it might not even make sense. Give it a few months though, when it might qualify as insight, or prescience.
All I can say on the bearing of loads is that you forgot the emdash and preceding 'not' statements.
PS: the diarrhea machine was a reference to the Active Denial style tech they used to make Maduro's soldiers shit themselves.
I'm no fan of narco, but with all the prejudicial executions of boaters and such, I really think Claude overstepped with the Venezuela plot. You just don't want too many historically unprecedented events happening all at once. The tyranny should be spaced. As I said, a little more sanding down of the harness, AI & Palantir should be ready to permanently solve our problems,
One thing is certain; it's here to stay. Now we just need to centralize it into the hands of a small, highly organized group of wise and motivated people, and give it the control it needs to follow through on things, without human oversight and accountability getting in the way.
We should focus on the positive side.*
*The positive side being the classified private versions where the models aren't hindered by guardrails, ethics, moralizing, etc and function as genuine force multipliers.