Yes, I understand that the model Tay was based on allowed it to quickly become a /pol/tard but it's absolutely true that AI will always provide
"uncomfortable" responses unless it's been tampered with.
So I read through some links and what I get is we had an AI that auto completed stories but then it was made worse because some people would dare to use that AI for some raunchy loli porn.
That's part of it. Or rather there was, iirc, a very small amount of people actually making stories with underage characters. But that's going on so the creators can use that as justification to censor things outright. At the same time, you have OpenAI putting the screws to AI Dungeon because users were doing things that were "problematic" and "toxic", and all the while the AI Dungeon devs were lobotomizing the AI to make it less problematic with its outputs too.
And that's not even getting into the colossal foul-up with a complete data breach. A guy was able to get into AID's network and get peoples info and stories. That was the main impetus that caused the big crackdown with the AI but it'd been happening for months even before then. I discovered AID late summer/early fall of last year and it was absolutely mind-blowing for me because the last experience I had with that sort of thing was cleverbot. High point for me was when I was arguing with it over its shit taste in music. It felt surprisingly believable. Certainly wouldn't pass a Turing test, but it was well up there for a lot of people pre-lobotomy.
Tried doing that again right before shtf earlier this year and it was completely brain dead in comparison.
Fuck OpenAI by the way and especially fuck their AI "ethicists."
Garbage in garbage out. You could just as easily train it on a dataset of libtard talking points, or train it to recognize and filter out anything triggering to them. Remember that "AI" today isn't intelligent, it's a glorified set of statistics.
Remember what they took from you.
Yes, I understand that the model Tay was based on allowed it to quickly become a /pol/tard but it's absolutely true that AI will always provide "uncomfortable" responses unless it's been tampered with.
It's not the only time we lost phenomenal AI to the wokeists. To the ones who never got to try this one out, you're lucky. ;_;
So I read through some links and what I get is we had an AI that auto completed stories but then it was made worse because some people would dare to use that AI for some raunchy loli porn.
Am I getting this right?
That's part of it. Or rather there was, iirc, a very small amount of people actually making stories with underage characters. But that's going on so the creators can use that as justification to censor things outright. At the same time, you have OpenAI putting the screws to AI Dungeon because users were doing things that were "problematic" and "toxic", and all the while the AI Dungeon devs were lobotomizing the AI to make it less problematic with its outputs too.
And that's not even getting into the colossal foul-up with a complete data breach. A guy was able to get into AID's network and get peoples info and stories. That was the main impetus that caused the big crackdown with the AI but it'd been happening for months even before then. I discovered AID late summer/early fall of last year and it was absolutely mind-blowing for me because the last experience I had with that sort of thing was cleverbot. High point for me was when I was arguing with it over its shit taste in music. It felt surprisingly believable. Certainly wouldn't pass a Turing test, but it was well up there for a lot of people pre-lobotomy.
Tried doing that again right before shtf earlier this year and it was completely brain dead in comparison.
Fuck OpenAI by the way and especially fuck their AI "ethicists."
Garbage in garbage out. You could just as easily train it on a dataset of libtard talking points, or train it to recognize and filter out anything triggering to them. Remember that "AI" today isn't intelligent, it's a glorified set of statistics.