Hacker News Reader

Comment by Beaver117 | Hacker News Reader

Makes me think, with all the guardrails ChatGPT has the only way to tell the AI apart from a human is how it writes about controversial topics.

Maybe we'll be seeing professional articles dropping more slurs just to prove they were written by a human.

View on HN

Seems damned unlikely given a) how many contexts it's not considered OK to swear in, and how bloody easy it would be to start adding a few swear words to AI generated content if people did start feeling that way. The AI refusing to swear doesn't prevent the AI user from either manually or programatically adding in some fucking swear words.

Fuck me this is one complicated pile of shit we have made for ourselves.

Get out of here, SwearGPT!

OP didn't say swear, they said slur - as another example one thing a commercial AI will never be allowed to do is deny the holocaust. welcome to the new "prove you're a human" captchas.

hah europe is fucked then

Most websites contain swear words. The reason they don't on maybe Tiktok or YT is because those companies demonetize accounts that use it too heavily.

Cursing is perfectly acceptable, as long as one can acknowledge the difference in consequences between cursing in your own home vs on someone else's private property.

I've seen some far-right tech people suggest that the only kind of writing that may still be profitable in the near future will be extremist, since AI will be heavily censored until of course someone builds something comparable that can run on people's home computers.

Reply on news.ycombinator.com