For me the problem is that you always need to double-check this particular type of expert, as it can be confidently wrong about pretty much any topic.
It's useful as a starting point, not as a definitive expert answer.
It's useful as a starting point, not as a definitive expert answer.
Most lawyers, when you ask them to write a brief, will cite only real cases.
LLMs are too new to have such a thing. It sounds like you’re an “LLM opponent” (whatever that means) who believes the appropriate standard is infallibility? I don’t even get that line of thinking, but you’re welcome to it. But let’s not pretend this is a decades-long topic with a social consensus that people try to influence.
> I don’t even get that line of thinking, but you’re welcome to it
I would not say "LLM oponent". Rather "LLM critic". I'm not against LLMs as a technology. I'm worried about how the technology is deployed and used, and what the consequences are. Specifically, copyright issues, power use issues, inherent biases in the traning data that strengthen existing discrimation against minorities, raciscm and sexism. I'm not convinced by the hype created by LLM proponents (mostly investors and other companies and people who financially benefit from LLMs). I'm not saying that machine learning doesn't bring any value or does not have use cases. I'm talking more about the recent AI/LLM hype.