3 ms·
Today I discovered that even if I give the AI a safe prompt 1 time in 4 it decided to generate some "bad" stuff, it tells me it can't comply but it takes my cre
by simion314 4y ago
Today I discovered that even if I give the AI a safe prompt 1 time in 4 it decided to generate some "bad" stuff, it tells me it can't comply but it takes my credits. This in an API call, why not use their fucking safe shit and put the robot to work over and over again to create safe output?
Bastards, I suspect the issue was that the prompt contained the word "monkey" and I suspect since this is an USA company they put tons of safe filters in, from all the big groups(conservatives, religious, liberals ) so nobody gets offended and cause a PR issue.
TL:DR they are filtering what we can do, and when they do mistakes we still pay for it.
- tough 4y agoAt least stability won't charge you for rugged outputs
- bpodgursky 4y agoWere you doing anything that even incidentally compared a human to a monkey? Yeah, that's going to hit some red flags on the algorithmic bias bits.
- TeMPOraL 4y agoEven though humans are just a barrel of filthy monkeys, speeding through the void of space with no idea what they're doing? I guess we won't be having discussions with the AI about "Expeditionary Force" book series.
- simion314 4y agoNo, at least not intentionally. And as I said the input past the moderation checks we do before calling the complete APIs. And the respone was most of the time valid but soemtimes it would be moderated. The stupid shit is that there is no flag for us to know is moderated , and the chatbot response with a similar message but in different wording, so you need to guess that the response was moderated.