...Experts Warn The Odds Of An AI Catastrophe Could Be As High As One In Ten
#31
(06-15-2018, 06:40 PM)ServusDei Wrote: Computers that have "taught" themselves face recognition have first been fed a bunch of real faces, thatway the computer can put in its memory that all faces are roundish, have dark spots (nose, eyes, mouth) and in specific locations, etc.

So the programmers have basically taught the computer how to "learn," and the computer has just "learned" how to find similarities between images and what it has been "taught" was a face.

That is how it usually works.  The programmers intend to teach the computers facial recognition and feed it data for it to use for this task.

No, this was accomplished by computers that were not fed a bunch of real faces.

The computer simply learned to recognize the faces of its own programmers without being programmed to do so.

I'm trying to find the link and I'll post it if I can.
Reply
#32
Can't find the link, but I'll keep looking.

Anyway, the real fun begins when humans become chipped as part of the hive mind.   :s

Reply
#33
This short vid explains how a robot which "was not programmed to move" taught itself how to walk:

Reply
#34
(06-15-2018, 06:27 PM)ServusDei Wrote: I did. A computer can never do what the programmer did not program it to do. So if a programmer programmed it do do that, then it is possible, but computers cannot do that on their own.

At the 35 sec mark this US Defense Expert defines 

Artificial Intelligence as, "Software which writes itself. It writes its own updates." 

 He goes on to say that after it has written its own updates, you can't take it apart again and figure out what it has done.

Reply
#35
Artificial intelligence can be programmed to update itself with some random new things that it determines, but it was programmed to do that, and it is possible to make it log everything it does.
:monstrance:Deo Gratias et Ave Maria! :monstrance:
Pray the Rosary

A Dieu mon ame,
Mon arme au roi,
Mon Coeur a la dame,
Mon honneur a moi!
Reply
#36
(06-18-2018, 12:31 PM)ServusDei Wrote: Artificial intelligence can be programmed to update itself with some random new things that it determines, but it was programmed to do that, and it is possible to make it log everything it does.

How do you explain the video of the robot which was not programmed to move at all and had no information about its own size, shape or makeup which taught itself how to walk?
Reply
#37
It was programmed to experiment with commands to its moveable engines, and it got it right, after some practice.
:monstrance:Deo Gratias et Ave Maria! :monstrance:
Pray the Rosary

A Dieu mon ame,
Mon arme au roi,
Mon Coeur a la dame,
Mon honneur a moi!
Reply
#38
(06-18-2018, 01:06 PM)ServusDei Wrote: It was programmed to experiment with commands to its moveable engines, and it got it right, after some practice.

I see.  Well, I'm no expert by any means on computer stuff.

But how do you account for all the warnings from the experts who work with AI?

For instance there was the robot who aiming to shoot at the crowd and was tackled and stopped by a marine described in the Jay Tuck video above.
Reply
#39
Well, if computers are programmed to try out random things, one never knows what might lead to something bad. That's why robots should never be allowed to experiment with weapons unless in a totally sealed location.
:monstrance:Deo Gratias et Ave Maria! :monstrance:
Pray the Rosary

A Dieu mon ame,
Mon arme au roi,
Mon Coeur a la dame,
Mon honneur a moi!
Reply
#40
(06-18-2018, 01:13 PM)ServusDei Wrote: Well, if computers are programmed to try out random things, one never knows what might lead to something bad. That's why robots should never be allowed to experiment with weapons unless in a totally sealed location.

Well, I'm sure the powers that be would never let anything bad happen.  :dodgy:
Reply




Users browsing this thread: 1 Guest(s)