I hope OP is still in a drawing mood.
>>103990
That is maybe not entirely fair. LLMs do often make stuff up, but when pressed normally admit it and, bizarrely, apologize. I have these moods where I want to ask questions about, for example, quantum mechanics as it relates to cosmology (it's a compulsion, don't @ me), or nucular reactor stuff (had to install models locally for that one, because both Google and Duckduckgo's AIs clammed up and wouldn't tell me anything interesting). I think that they can be useful information-gathering and information-summarizing tools so long as we NEVAR 4GET how often they make up stuff. I have never seen two LLMs hallucinate exactly the same stuff, so ask several different models your questions for confirmation and clarification.
I don't claim to know what the future holds. The CCP is absolutely convinced with a fanatical fervor that AI will Very Soon Now be a wonderful bottomless cornucopia of novel, never-before-conceived-of new weapons to use against the West. I don't know whether AI is going to do any of this for them or just hallucinate nonsense at them until they bankrupt themselves, but they can't build datacenters fast enough. They can't write trillion-dollar checks fast enough. The Russian government is equally enthusiastic but doesn't command the resources to make a real go at any of this and knows it. And none of that is going to get "legislated away," Eliezer Yudkowski's horrifying Dr. Strangelove fever dreams of a nuclear-armed United Nations bombing nations that don't ban AI notwithstanding.
What were we talking about again?
