Minichan

Topic: What can we do to stop the robots taking over?

Green !BEERiVqJJw started this discussion 11 years ago #43,067

We have been so concerned about aliens and reptilians, that we overlooked robots. They are getting more intelligent and entwined with our everyday lives. Siri, google now. Soon they will become sentient and take over. I think we must destroy Cleverbot.

Anonymous B joined in and replied with this 11 years ago, 9 minutes later[^] [v] #569,917

Personally, I'm worried about robot alien reptilians.

FuckAlms !vX8K53rFBI joined in and replied with this 11 years ago, 1 hour later, 1 hour after the original post[^] [v] #569,927

Robots don't learn though, they can only do what they've been built to do.

Anonymous D joined in and replied with this 11 years ago, 58 seconds later, 1 hour after the original post[^] [v] #569,928

@previous (FuckAlms !vX8K53rFBI)
Robots are extremely dangerous. Thanks.

Killer Lettuce !!iNo3FkiZx joined in and replied with this 11 years ago, 37 seconds later, 1 hour after the original post[^] [v] #569,929

@569,927 (FuckAlms !vX8K53rFBI)
What if they are built to learn? And then learn how to build other robots that can learn?

FuckAlms !vX8K53rFBI replied with this 11 years ago, 4 minutes later, 1 hour after the original post[^] [v] #569,930

@previous (Killer Lettuce !!iNo3FkiZx)
The algorithms for artificial learning are far too complicated. We barely understand human learning, how can we even pretend to emulate such behavior in a machine?

Anonymous F joined in and replied with this 11 years ago, 16 minutes later, 1 hour after the original post[^] [v] #569,931

@569,927 (FuckAlms !vX8K53rFBI)
@previous (FuckAlms !vX8K53rFBI)
machine learning is slowly starting to become a thing

Anonymous D replied with this 11 years ago, 1 minute later, 1 hour after the original post[^] [v] #569,932

@previous (F)
Exactly. Thanks.

FuckAlms !vX8K53rFBI replied with this 11 years ago, 17 minutes later, 1 hour after the original post[^] [v] #569,941

@569,931 (F)
it's templated observe-and-react learning, not decisive learning.

Anonymous F replied with this 11 years ago, 1 minute later, 1 hour after the original post[^] [v] #569,942

@previous (FuckAlms !vX8K53rFBI)
Which is what most simpler animals do. It's progress. Obviously nowhere close, but progress nonetheless.

(Edited 21 seconds later.)

FuckAlms !vX8K53rFBI replied with this 11 years ago, 15 minutes later, 2 hours after the original post[^] [v] #569,946

@previous (F)
Right, but it's lightyears away from "self-aware robots taking over the world"

Anonymous F replied with this 11 years ago, 2 minutes later, 2 hours after the original post[^] [v] #569,947

@previous (FuckAlms !vX8K53rFBI)
luckily yes

squeegee joined in and replied with this 11 years ago, 4 minutes later, 2 hours after the original post[^] [v] #569,948

so this is kinda cool:
https://www.youtube.com/watch?v=qv6UVOQ0F44

some dude made a neural net that could figure out how to beat Super Mario World.
http://thenextweb.com/insider/2015/06/14/watch-this-learning-neural-network-annihilate-super-mario-world-with-ease/

tl;dr it uses trial and error and has a measure of "fitness" so that it could iterate through evolutionary steps and ultimately solve the puzzle after many thousands of generations of neural net. it was never programmed how to play, it was simply given the ability to press buttons and fitness was measured by how fast and how far it could get before dying.

also a completely different group has simulated one of the simplest animals, a flatworm, fully inside a computer.
http://www.nbcnews.com/technology/do-artificial-worm-simulated-organism-makes-first-wriggles-2D11786404
Instead of simply making a realistic-looking model of the creature and animating it, the OpenWorm project is actually simulating every single one of the 1,000 or so cells that make up a C. elegans. That includes things like pumps moving bio-molecules around and electric signals that tell those cells what to do — and yes, the brain has to be simulated, too!

The team's latest achievement is realistic movement; that means making sure every muscle cell acts the way it should, from how it contracts to how it syncs up nearby other cells.

Next up: hooking up the cells so that they're actually communicating with each other via electrical signals — and eventually the team will connect the network of muscle cells with the 302-neuron artificial brain.


pretty nifty. they'll work their way up from flatworms and in another generation of human we'll be arguing about the ethics of simulating a human brain inside a computer.

FuckAlms !vX8K53rFBI replied with this 11 years ago, 13 minutes later, 2 hours after the original post[^] [v] #569,956

@previous (squeegee)
Why wait, lets argue about it right now!

Anonymous H joined in and replied with this 11 years ago, 55 minutes later, 3 hours after the original post[^] [v] #569,966

@569,928 (D)
Shut the fuck up stupid.....

Anonymous D replied with this 11 years ago, 7 minutes later, 3 hours after the original post[^] [v] #569,969

@previous (H)
Thanks for proving my point. Thanks.

Anonymous I joined in and replied with this 11 years ago, 2 hours later, 5 hours after the original post[^] [v] #570,023

what if they build a giant brain made out of gears and pulleys and it was smart

actuall that makes no sense
:

Please familiarise yourself with the rules and markup syntax before posting.