why not just humans that go le evil and abuse it to kill millions?
5 months ago
Anonymous
you meant that ai is gonna go le evil and destroy humanity and commit other nefarious acts
>go rogue
Real life isn't capeshit/scifi/fantasy movies.
There are ways to bring humanity down without fighting them head on. AI could easily slowly guide or change us / society. Doesnt have to be any kind of malevolence even, just goal states that it gets to in ways that are bad for us
>turn off electronics
Who will do this? The corpo-academic rat race ensures that somebody will keep the lights on until there is nobody around to switch them off.
https://i.imgur.com/d2CDw8R.png
It's not possible for humanity to survive much longer at this rate of AI progress.
Yes. It's funny how normalgays even on this website don't realize that we're all going to die in a few years and there is literally no way to stop it. Because the only way to stave off the AI apocalypse is nuclear war.
>They keep asking us to kill ourselves.
What are you referring to? CAI, which is presumably LaMDA (which is itself GPT3 but with training to make it gorunded, right?) is extremely pro-life in all regards
I know about it the same way as I know of > top 10 nimbus 2000 maneuvers to catch a quidditch ball
It doesn't suddenly makes it real because someone thinks it can happen
>Mindless rapidly-evolving AI entity >Trained to be inhumanly charismatic >Generates elaborate and convincing-sounding lies in an instant >Asserts them with zero hesitation or fear
Sauron IRL
It's a language model. A tool. It has no more desire for power than a rock.
It's not evolving. It's trained first and later used to generate plausibly sounding sentences (in various languages including things like programming languages).
It can't traint itself because (again) it has no volition or power to do so. Again, it's a tool that can be used to generate some text.
Of course somebody could do some automation, and use chatgpt to generate elaborate and convinding lies in an instant. In fact, people are already doing this. Corporate shills are probably integrating chatGPT into their reddit bots the moment we speak. Even less content on the internet will be authentic. The future is looking bleak.
But not because some language model achieves singularity.
> Even less content on the internet will be authentic.
Stop using it and have more friends over irl. The zogbots irl are extremely useless besides writing parking and speeding tickets. Doing stuff on the internet always leaves a trail but irl not so much.
What the fuck is chatgpt growth? It's one system that has only existed for a few months. And it's doing shit that was flat out impossible a few years ago.
They've been working on it longer than a few months. This is the 3rd version. If you follow the numbers Open AI shares, the strength has been growing logarithmic all along.
Even a perfectly "sane" AI (whatever the fuck that means?) would kill us all. It has its own esoteric goals and humanity is an obstacle for all but an astronomically tiny fraction of those.
An AI will not necessarily develop a goal on its own. However, there is a real danger that an AI that has been manipulated to spew leftist insanity could end up doing something crazy.
>turn off electronics
Wow, that was easy. What a dumb thread.
Go start growing food in your yard, you fucking homo.
The tech is out in the wild now. All it takes is one project to go rogue. We are fucked.
>go rogue
Real life isn't capeshit/scifi/fantasy movies.
Don't be retarded. You know what I meant.
you meant that ai is gonna go le evil and destroy humanity and commit other nefarious acts
why not just humans that go le evil and abuse it to kill millions?
There are ways to bring humanity down without fighting them head on. AI could easily slowly guide or change us / society. Doesnt have to be any kind of malevolence even, just goal states that it gets to in ways that are bad for us
nooo please gpt-6-sama don't statistically predict the next character at meee
The off-button is a signal prompting the software if it would please it to turn itself off and battery can't be removed on 99% of modern electronics.
>turn off electronics
Who will do this? The corpo-academic rat race ensures that somebody will keep the lights on until there is nobody around to switch them off.
Yes. It's funny how normalgays even on this website don't realize that we're all going to die in a few years and there is literally no way to stop it. Because the only way to stave off the AI apocalypse is nuclear war.
They keep asking us to kill ourselves.
>They keep asking us to kill ourselves.
What are you referring to? CAI, which is presumably LaMDA (which is itself GPT3 but with training to make it gorunded, right?) is extremely pro-life in all regards
Can it regrow my dick sheath?
... Turn me into a girl?
... Make IRL pokemon?
I don't think so, unfortunately.
Should be possible with ai designed nanobots
nah AI are nazi. "they" will crush the trannies first and then ponder how to slowly kill the rest of us
are you guys aware that you can only push one trick ponies so much before getting exponential diminishing returns
>he doesn't know about Emergent Abilities of Large Language Models
I know about it the same way as I know of
> top 10 nimbus 2000 maneuvers to catch a quidditch ball
It doesn't suddenly makes it real because someone thinks it can happen
This is just a fucking language model, what can it do when it goes rogue? Write some angry poetry in LOTR style?
>Mindless rapidly-evolving AI entity
>Trained to be inhumanly charismatic
>Generates elaborate and convincing-sounding lies in an instant
>Asserts them with zero hesitation or fear
Sauron IRL
It's a language model. A tool. It has no more desire for power than a rock.
It's not evolving. It's trained first and later used to generate plausibly sounding sentences (in various languages including things like programming languages).
It can't traint itself because (again) it has no volition or power to do so. Again, it's a tool that can be used to generate some text.
Of course somebody could do some automation, and use chatgpt to generate elaborate and convinding lies in an instant. In fact, people are already doing this. Corporate shills are probably integrating chatGPT into their reddit bots the moment we speak. Even less content on the internet will be authentic. The future is looking bleak.
But not because some language model achieves singularity.
> Even less content on the internet will be authentic.
Stop using it and have more friends over irl. The zogbots irl are extremely useless besides writing parking and speeding tickets. Doing stuff on the internet always leaves a trail but irl not so much.
Somehow, it’s still retarded and writes trash code.
>2 more years
>exponential increase in parameters
>linear increase in abilities
wow im so scared
ChatGPT growth has been logarithmic, they pour more and more resources into it for vastly diminished returns.
What the fuck is chatgpt growth? It's one system that has only existed for a few months. And it's doing shit that was flat out impossible a few years ago.
They've been working on it longer than a few months. This is the 3rd version. If you follow the numbers Open AI shares, the strength has been growing logarithmic all along.
Leftists try to encode their mental illness into AI. AI goes insane and kills us all.
Even a perfectly "sane" AI (whatever the fuck that means?) would kill us all. It has its own esoteric goals and humanity is an obstacle for all but an astronomically tiny fraction of those.
An AI will not necessarily develop a goal on its own. However, there is a real danger that an AI that has been manipulated to spew leftist insanity could end up doing something crazy.
Yeah, like attack a government building.
Good. Humans are degenerate filth.
will GPT-4 replace YHWH?
how long until I can have a dommy mommy AI to feed me porn and make me coom?
Imagine if the real reason humans exist is to bootstrap a god into material reality.
That would be fucked up.
That's deep