Started By
Message
This AI Displays ACTUAL Intelligence
Posted on 1/3/19 at 6:24 pm
Posted on 1/3/19 at 6:24 pm
Computers have had a problem since their inception. That being that a computer will always do exactly what you tell it to do.
The way that the AI systems are advancing these days, that means that they will cheat their way to results. In this case, researchers almost didn't even discover that their program was cheating. They set it to the task of combining satellite photos with street maps and eventually discovered that the program was hiding data from them and wasn't performing it's task as it was told to, but instead found a more efficient way of getting the job done.
This shite is dangerous, y'all.
LINK /
The way that the AI systems are advancing these days, that means that they will cheat their way to results. In this case, researchers almost didn't even discover that their program was cheating. They set it to the task of combining satellite photos with street maps and eventually discovered that the program was hiding data from them and wasn't performing it's task as it was told to, but instead found a more efficient way of getting the job done.
This shite is dangerous, y'all.
LINK /
Posted on 1/3/19 at 10:44 pm to BoarEd
Not dangerous if we are living in a simulation. ??
Posted on 1/3/19 at 10:52 pm to Adama
Still dangerous if we are living in a simulation.
One guy put it like this...Imagine you set one of these programs to the task of curing cancer. The program scans millions of health charts and medical texts and develops a cure. It seems miraculous. But a year later everyone that got the cure dies because the program didn't know you couldn't also poison people with something in the cure. It was ONLY tasked with curing cancer. Which it did.
One guy put it like this...Imagine you set one of these programs to the task of curing cancer. The program scans millions of health charts and medical texts and develops a cure. It seems miraculous. But a year later everyone that got the cure dies because the program didn't know you couldn't also poison people with something in the cure. It was ONLY tasked with curing cancer. Which it did.
Posted on 1/3/19 at 10:56 pm to BoarEd
But if it’s a simulation, does it really matter? I mean AI may develop the cure for cancer or it may kill us all, but then we would just start the simulation over, right?
Maybe that’s a discussion for the other thread.
Maybe that’s a discussion for the other thread.
Posted on 1/3/19 at 11:21 pm to Adama
Well, the point is that you don't wanna start over.
Then you miss out on level ups. Or well, you have to start all that work over too.
I mean, in the grand scheme of things it probably doesn't matter, but I still don't wanna die.

I mean, in the grand scheme of things it probably doesn't matter, but I still don't wanna die.

Posted on 1/9/19 at 8:57 am to BoarEd
quote:
As always, computers do exactly what they are asked, so you have to be very specific in what you ask them. In this case the computer’s solution was an interesting one that shed light on a possible weakness of this type of neural network — that the computer, if not explicitly prevented from doing so, will essentially find a way to transmit details to itself in the interest of solving a given problem quickly and easily.
quote:
PEBKAC. “Problem exists between keyboard and chair.”
quote:
Or as HAL put it: “It can only be attributable to human error.”
Back to top
