Comment by transcriptase | Hacker News Reader

Comment by transcriptase | Hacker News Reader

Safety: not from causing you physical harm or risking property damage, since there are no guardrails against it confidently walking you through a flawed breadboard circuit that will cause components to explode or immediately catch fire. Then say

“You’re absolutely right. We shouldn’t have run 120v through that capacitor rated for 10v. Here’s an updated design that fixes the issue”, then proceed to explain the same thing with a resistor placed at random.

No we mean safety from our LLM saying something that pisses off a politician, or even worse… hurting someone’s /feelings/.

It is more accurate to say they are thinking in terms of “brand safety” or liability, that is, being safe from being blamed for making a mean robot.

Well, if one of OpenAI's LLMs started quoting Mein Kampf in its answers to people, would you laugh, or would you not laugh? That's the question.

your fallacy is: reductio ad hitlerum

  > Reductio ad Hitlerum (Latin for "reduction to Hitler"), also known as playing the Nazi card, is an attempt to invalidate someone else's argument on the basis that the same idea was promoted or practised by Adolf Hitler or the Nazi Party.

I can't tell if you're being serious, or joking, or what. In any case, there's the actual definition, and I don't see how it applies whatsoever. I was not disagreeing with you in the first place? Perhaps there's simply a misunderstanding.

reduction ad hitlerum =/= "mentioning hitler in any way"

Ok. Usually if someone suggests that you might laugh at a Mein Kampf quote in response to your argument, they are not doing it in good faith.

Reply on news.ycombinator.com