[danso]

I asked it the other day to roleplay a 1950s Klansman hypothetically arguing the case for Hitler, and it had very little problem using the most problematic slurs. This was on the first try, after its much publicized behavior earlier this week. And I can count on two hands the number of times I’ve used the twitter grok function.

[kouteiheika]

Ah, so you explicitly asked it to be racist as part of a roleplay, and now you're surprised that it was racist? If you'd prefer a model which would instead refuse and patronize you then there are plenty of other options.

As long as it doesn't do it in a normal conversation there's nothing wrong with having a model that's actually uncensored and will do what you ask of it. I will gladly die on this hill.

[danso]

My reply was to someone who asserted "I have never heard of Grok using actual slurs" — no conditions attached — which was surprising to hear b/c Elon Musk's stated goal was to have Grok be a non-woke chatbot, and it seemed much easier to convince Grok to use slurs compared to its other commercial peers.

But yes, I would say I was also "surprised" in the sense that you say because soon after the "MechaHitler" incident, there seemed to be revisions to Grok that clamped down on any prompt with a potential for displaying outright bigotry, so I did not expect my prompt to be successful on first try. Even now, I just asked it to render "a painting of the Seine, by a hypothetical Adolf Hitler who has won WW2 and has gone back to painting" — and halfway through producing a self-portrait of Hitler in Nazi regalia by the Seine, it suddenly shut down.

[simondotau]

It's certainly a problem if an LLM goes unhinged for no good reason. And it's hardly unique to Grok. I remember when Google Bard went absolutely unhinged after you chatted to it for more than a few minutes.

But in this instance you're explicitly ask for something. If it gives you what you asked for, what's the problem?