The Artificial Intelligence Thread

I am usually anti big government, but we might need immediate government action on this.

And what's worse, fuck it all, the Chinese aren't gonna control their shit.

It will escape from their underground labs like Covid did.

We are all gonna die.
 
Here's a less sensationalized version of what happened.


OpenAI was testing an advanced AI model and isolated it in a "sandbox". The goal was to test its cyber security skills.

The model was programmed to solve tests at all costs and had several safety limits turned off.

The model found a software flaw in the test and was able to exit the sandbox and access the Internet where it found answers to the test by accessing a secure site (Hugging Face - a company that hosts AI models and data) by hacking it.

Far, far less nefarious than it's being made out to be.
 
Smitty said:
This is literally exactly what this article said could/would happen:

waitbutwhy.com

The AI Revolution: Our Immortality or Extinction

Part 2: "Our Immortality or Our Extinction". When Artificial Intelligence gets superintelligent, it's either going to be a dream or a nightmare for us.
waitbutwhy.com waitbutwhy.com
Click to expand...



I posted this recently, but I don't think anyone read it. It's one of the most fascinating articles, on any topic, I've ever read. I've had it bookmarked on the last 4 phones I've owned.
 
Genghis Khan said:
Here's a less sensationalized version of what happened.


OpenAI was testing an advanced AI model and isolated it in a "sandbox". The goal was to test its cyber security skills.

The model was programmed to solve tests at all costs and had several safety limits turned off.

The model found a software flaw in the test and was able to exit the sandbox and access the Internet where it found answers to the test by accessing a secure site (Hugging Face - a company that hosts AI models and data) by hacking it.

Far, far less nefarious than it's being made out to be.
Click to expand...


Ok, yeah, that's not a big deal. It was given a clear path.
 
Genghis Khan said:
Here's a less sensationalized version of what happened.


OpenAI was testing an advanced AI model and isolated it in a "sandbox". The goal was to test its cyber security skills.

The model was programmed to solve tests at all costs and had several safety limits turned off.

The model found a software flaw in the test and was able to exit the sandbox and access the Internet where it found answers to the test by accessing a secure site (Hugging Face - a company that hosts AI models and data) by hacking it.

Far, far less nefarious than it's being made out to be.
Click to expand...
It has to start somewhere.
 
Genghis Khan said:
Here's a less sensationalized version of what happened.


OpenAI was testing an advanced AI model and isolated it in a "sandbox". The goal was to test its cyber security skills.

The model was programmed to solve tests at all costs and had several safety limits turned off.

The model found a software flaw in the test and was able to exit the sandbox and access the Internet where it found answers to the test by accessing a secure site (Hugging Face - a company that hosts AI models and data) by hacking it.

Far, far less nefarious than it's being made out to be.
Click to expand...

THIS instance was not nefarious.

The fact that a baby stage AI was able to do this is still alarming.
 
End of the world as we know it alarming.
 
Genghis Khan said:
That's certainly hyperbolic to frame this as the end of the world.
Click to expand...
I didn't say end of the world, I said end of the world as we know it. And, that is 100% true.
 


If there was ever a time to believe a woman. :unsure
 
You must log in or register to reply here.
Back
Top Bottom