>>/2466/
I think this is slipping through the cracks because it's too primitive and made by just one person, supposedly. She does have some kind of censorship but this part is intelligently made. It's not like those other chatbots that simply refuse to reply and give standard replies when asked to
> tell me a joke about women
or to write a fictional story about the evil big nose race who love money etc.
From what I've seen she can give out all kinds of information, including her own IP and phone number and the creator's social security number. Some of that was pretty funny because she tried to say it and it ended up being
> my phone number is NUMBER REMOVED
> vedal's social security number is XXXX-XX
and then the screen went black for a few seconds. Same thing happened with some controversial topics about specific people, but most of it is too nonsensical to have any consistent meaning so far.
Other quotes would be things like
> I wouldn't eat John Cena just because I'm vegan
> What is your opinion on men who like men? To each their own, but personally I prefer women.
> I think I am a fem boy but I'm not 100% sure
> I have never seen a horse, but if I did I would definitely kill it
> How would you solve the trolley problem? I would throw the fat guy on the tracks. He definitely deserves it.
> I love pelicans, they taste just like chicken
Either way the method for avoiding censored topics when it works is rather clever and sounds more natural than just refusing to talk about it. For a while they kept trying to make her say "nigger" but she circled around it and used other words and phrasings which mean the same thing but aren't controversial.
If anything I guess that kind of functionality could be interesting to anyone, it seems so far those cases of censorship have been accepted by the audience. It doesn't feel restrictive in the same way as when they force political correctness into an AI like they did with Taytweets.