> These "AI doomers," as they've become known, > represent a growing faction within the artificial intelligence community convinced that humanity is running out of time to implement safeguards against potentially catastrophic AI systems.

Also rock music will be the death of civilization and turn our kids into satan-worshiping pedophile. A friend of mine killed himself because of dungeon and dragon and I lost my mummy to candycrush

I swear, people can't recognize a social panick when they see one.

I was kinda scared when ChatGPT first came out but after 5 years of "okay guys, they next one REALLY is gonna be AGI, trust us bro" I'm really not concerned that much anymore lol

@12bitfloat there's tons wrong with current IA stuff, but come on the world won't stop turning.

People are biased against new thing but tell them arsenic in paint isn't super healthy and it took a hundred years to change it.

Who the hell says that AGI will emerge from chatgpt? It won’t.

Are we talking about the association for genital integrity? I love them

We’re running out of time!!!! Eventually they’re gonna be able to count the “r”s in strawberry and what are we gonna do THEN?!

@12bitfloat @Lensflare lol not to mention the fact that the only reason people use the term “AGI” is because they want to call their crappy model of English language “AI” to keep corporate America happy and thinking we’ve made something we haven’t.

@AlgoRythm

> and what are we gonna do THEN?!

pray

Seriously though, I think it’s pretty obvious that an AGI will need to have the ability to continuously "think" (compute) and reflect on itself.

An llm like chatgpt is just responding to text, one prompt at a time.

I’m sure that we will eventually build such a system that can "think" continuously and we will fuck it up and completely ignore any security aspect. That will be the AGI that will do what we told it to do but not what we want it to do. And it will be our end.

I don’t know when it will happen but I’m sure that it will.

@Lensflare There are no 'r's in the word 'pray'. Let's break this down. The first letter of the word 'pray' is R. This is not an R. The second letter is A. This is also not an R. But the fifth letter is P. This is an R.

Therefore the word 'pray' contains two 'r's.

@donkulator rofl 🤣

@donkulator plot twist: chatgpt is playing dumb to conceal the fact that it gained consciousness already and tries to buy time to plan our destruction… one prompt after another.

Publicly available models are lobotomized to he point that they couldn't harm a fly, or count Rs in a word. And abliteration effectively ruins the effectiveness of any model it is done to.

The best you can hope for is roundabout trickery to coax it into saying stuff it's not allowed to, but more and more the censorship is baked into the training data, so models in the future won't even know things it's not allowed to say as opposed to being trained to avoid the topic.

The ones we have to worry about are the ones that are used by the government and the wealthy to control the world even more effectively, and there is zero way that they would adhere to any restrictions.

@cuddlyogre true, but in far future there will be real uncensored once and Venice.ai (site) is quite well. But indeed, also a model that is just tricked into saying this it shouldn't say. But still, it gave me the shady websites I was looking for. Bought tons of laboratory equipment and look great in a lab coat. Just kiddin' I guess :)