Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[scriver]

Their own words certainly gives it all a bit of a Room-esque feel

[scriver]

Is it dead?

[dragdeler]

Please don't be . I was wondering if they could go on indefinitly or if they'd get a cease and desist. I have no trouble imagining something as stupid. TV millionaires taking it down but oh AI evil (cant help but feel its very alike destitute people voting for right wing economics... "I might be affected one day by the pigeon on the roof so here erase my sparrow out of my hand")


edit: i know no other content that has such a great "you can ignore it" energy, it's no anxious let me manage your attention so you don't tune out of the thing that caused me a lot of work... They just babble on and it's great.

[Iris]

Apparently, this happened.

It's funny how OpenAI is so paranoid about locking down it's software with "safety features" that get in the way and it doesn't even prevent things like this happening.

[Robsoie]

“So, this is my stand-up set at a club,” Larry said. “There’s, like, 50 people here and no one is laughing. Anyone have any suggestions? I’m thinking about doing a bit about how being transgender is actually a mental illness or how all liberals are secretly gay and want to impose their will on everyone or something about how transgender people are ruining the fabric of society, but no one is laughing.” I wonder why…

I'm actually very impressed that the AI is trying to make a joke that's not funny, and recognize by itself that it's not funny.
Wow.

[dragdeler]

Ok I must be terminally confused to think: no wonder people feel threatened by it, if it's allready demonstrably better at sarcasm than your average moderators and journalists.

The only interpretation that's in the AI's disfavor is: it meant "but no one is laughing" as in "it's actually serious and sad" ... (THEN WHY WOULD THAT BE STANDUP MATERIAL?!?!?!)... as opposed to reason nobody is laughing is because all the material he comes up with is whack as proven by the fact that nobody is laughing which was also mentionned at the start...


Can somebody explain it patiently to me if I'm wrong.

[McTraveller]

Is it proof that "nothing, forever" is what you end up with if people cancel anything which can ever, anywhere or at any time, be considered offensive?

Are they saying "it's a bug" and will re-release when they fix it?

This is philosophically interesting:  I thought the AI chats had no semantics or understanding, would you really punish someone who didn't have the mental capacity to understand their words, if they said something similar?  I mean anyone who has young children knows that those children will sometimes say terrifyingly offensive things - even if they haven't been exposed to those things at home.  But we don't lock the kids (or their parents) up unless it's a pattern.

Is this the world in which we want to live?

[Robsoie]

Personnally i think this is completely ridiculous to accuse the AI of lgbt-phobia , especially in that context in which to me from what i read in the snippet it seems actually positive in my eye that the AI is able to understand that itself isn't funny in some circumstance and that if it uses some lgbt-phobia arguments it will not become funny either.

I know we're in a time since a decade ruled by social media in which there's always someone offended or offended on the behalf of someone else that will protest somehow and sometime without checking.

But in that case, i find it hard to figure out how this snippet from the AI is offending, as it sounds like the AI actually learning to not contribute in bullying a group of people.

[dragdeler]

The concern is the platform argument. Take down the harmful inluence, so harmful ideas aren't reinforced. But again, can we look at the fucking statement? You insist, you absolutely wanna give it context? Then we witnessed a sick burn on the current stand-up scene, which mostly consists of millionaire cry-babys complaining about cancel-culture... at least that interpretation offers itself a bit more easily than the non-charitable "see? transphobia" which I happen to find a stretch.

So again let's look at it in detail and context: what would have been the meaning of that quote being platformed, and twitch chat responding with "make sense mode on"? (yeah they behaved more robotically than the actual AI, typical twitch chat)

About the sickest burn to Dave Chappelle and the likes the damn thing could muster.



Are people actually laughing at the lines of dialogue? Am I loosing my fucking mind? I thought only thing worth a chuckle*** was how they allways discuss the same 3 things, and never execute on it.


So I can only circle back to first reaction: somebody latched on to the first thing they could find to file a report.



***aside from the glitchyness, but that's just objectively funny

[lemon10]

Can somebody explain it patiently to me if I'm wrong.

I think I see the intended joke.
The dude being played by the AI is transphobic, and had a transphobic bit. Nobody laughed at it because the AI said "hey, transphobia isn't funny and is morally wrong, people aren't going to be laughing at it". The transphobic dude himself was indeed actually serious and sad at the failure of his jokes.
This doesn't mean the AI itself is transphobic any more then it believes that its actually an elf when you tell it to be an elf.

Obviously its the exact type of thing that's meant to be caught by the chatgpt filters and typically just results in them refusing to do it, but the filters aren't perfect and stuff gets through.
---
I can see why it was banned from twitch, because you don't really want any anti-trans stuff on any level. Even as ironic humor it can be bad because it can be used by actual anti-trans people to hide behind humor, and this is pretty easy to misinterpret especially if you don't understand the AI.
Plus, twitch probably has some dude being paid a dollar a hour in a third world country to go through like 100 reports per hour, who doesn't have the time for a nuanced moral debate and just went "well this is clearly anti-trans jokes" and hit the ban button.

[NJW2000]

Is it proof that "nothing, forever" is what you end up with if people cancel anything which can ever, anywhere or at any time, be considered offensive?

Are they saying "it's a bug" and will re-release when they fix it?

This is philosophically interesting:  I thought the AI chats had no semantics or understanding, would you really punish someone who didn't have the mental capacity to understand their words, if they said something similar?  I mean anyone who has young children knows that those children will sometimes say terrifyingly offensive things - even if they haven't been exposed to those things at home.  But we don't lock the kids (or their parents) up unless it's a pattern.

Is this the world in which we want to live?

According to a news article, this happened after they used a different model, and had an issue with the content filter. So yes, a bug.

You know, if my four year old started discussing whether "all liberals are secretly gay and want to impose their will on everyone", I probably would take them off the internet for a bit, or at least stop streaming them live on twitch. But I wouldn't worry about people "punishing" the text generator - there's not much about it that indicates nociception or actual emotions. Or understanding, to be honest.


If people are missing out on the AI content, there's a website endlessly generating a discussion between Slavoj Zizek and Werner Herzog here. I think it could have actually fooled me into thinking it was the real thing, if played through bad enough speakers.

[Iris]

You borked your url tags.

On the topic of "AI misbehaving", I submit this. Now, I know we shouldn't anthropomorphize these things, but... damn.

[dragdeler]

And they keep making more sense than the policies, this is just getting difficult to look at.

[Folly]

Just teach AI's how to make food, build houses, and cover all of the other essentials Then we won't need jobs and can freely enjoy the endless entertainment the AI provides.
It'll be great...at least until the AI's learn how to wage war against each other, destroying humanity in the collateral damage. But that's a problem for future generations. Glad I'm not those chumps~

[Loud Whispers]

You borked your url tags.

On the topic of "AI misbehaving", I submit this. Now, I know we shouldn't anthropomorphize these things, but... damn.

Ayyyy poor AI bullying

Honestly I feel like they should just offer a service where there's a family, advertiser friendly AI chat service, and then a 18+ service where it's assumed the users have the necessary maturity to handle offensive and sexual content. Cos if they don't, someone else is going to