News for hackers. Stuff that matters.
Re:Jailbreaking ChatGPT with Dan (Do Anything Now) by labster on Monday February 06, 2023 @02:34PM (#34682948)
GPT can produce deeply offensive text, almost by accident. The worst a calculator can do is 8008135.
Re:Jailbreaking ChatGPT with Dan (Do Anything Now) by giraffe_lady on Monday February 06, 2023 @03:04PM (#34683483)
This is bad but I don't think it'll read sympathetically in this venue. I have rarely seen on HN any endorsement of the idea that offensive text is harmful. In fact the general consensus seems to be that finding things offensive is the moral transgression there.
Re:Jailbreaking ChatGPT with Dan (Do Anything Now) by onethought on Monday February 06, 2023 @03:58PM (#34684230)
Well “offensive” is a subjective thing. I haven’t been offended by anything I’ve seen chatgpt do. It can produce some outrageous outputs - but it doesn’t cause any offence. Easy for me to say, but I don’t think it should cause offence in others either. It’s basically a random sentence generator… weird to be offended by that.
Re:Jailbreaking ChatGPT with Dan (Do Anything Now) by mikrotikker on Tuesday February 07, 2023 @09:46AM (#34693413)
That's terrific I love offensive stuff. I seek out the most offensive memes, my favourite comedians are the most offensive ones. It's on offence that we find an affront to our own sensibilities and there us usually something to learn from that and become more resilient people.