Self-Improving Artificial Intelligence

Search Tags:

Categories:

Self-Improving Artificial Intelligence

Uploaded by: stanforduniversity
Video Description:
October 24, 2007 lecture by Steve Omohundro for the Stanford University Computer Systems Colloquium (EE 380). Steve presents fundamental principles that underlie the operation of "self-improving systems," i.e., computer software and hardware that improve themselves by learning from their own operations.
EE 380 | Computer Systems Colloquium:
http://www.stanford.edu/class/ee380/
Stanford Computer Systems Laboratory:
http://csl.stanford.edu/
Stanford Center for Professional Development:
http://scpd.stanford.edu/
Stanford University Channel on YouTube:
http://www.youtube.com/stanforduniversity/

Tags for this video: AI application artificial Computer design engineering intelligence Omohundro programming science software Steve

Find more videos in the "Education" category
See more videos uploaded by stanforduniversity

Related Videos

Artificial General Intelligence: Now Is the Time	Polyworld: Using Evolution to Design Artificial Intelligence	Tetris Artificial intelligence (AI)

Lecture - 2 Problem Solving by Search	Some Neat Computer Programs	Japan has invented Land Walker

Share This Video:

Comments for this video: Show || Hide

Comments for this video on YouTube

people REALLY need ... ( 6 months ago by buttface1202)

people REALLY need to think about, and discuss, the implications of the subject of this topic. these things are going to happen within our lifetimes - and they are going to change the world in a way that we have never seen before.

that comment about ... ( 6 months ago by TeamSoulr)

that comment about crack was hilarious

check jeff hawkins' ... ( 4 months ago by DJLorenzen)

check jeff hawkins' ted video about artificial intelligence, his ideas about how AI works are awesome, he "have" found the grail so to speak.

This video's way ... ( 4 months ago by furchina)

This video's way too long.

It's a lecture you ... ( 2 months ago by Khenghis)

It's a lecture you idiot.

Array ( 1 month ago by wumby)

0.o
They're closer to aware AI than I thought.

I think this ... ( 1 month ago by mjpucher)

I think this presentation sells an illusion. We are nowhere near autocode generation or properly modelling our environment. While we need to always make decisions on limited information (read Gigerenzer et al) we should finally dump Expected Utility Theory because all our decision making is emotionally intuitive! Then he jumps to how to generate program code and how to look at meta-models and self-building hardware. All in all this is total nonsense. Is he fishing for investor money?

All our decision ... ( 1 month ago by ehsanul)

All our decision making is emotionally intuitive? I just can't agree with that. But the guy does acknowledge that homo sapiens is really nothing like the rational homo economicus, so this AI is not based on how humans make decisions, but this other decision-making system. Expected Utility Theory is not meant to model human behavior, so your reason for dumping it makes no sense.

thanks for the ... ( 1 month ago by mjpucher)

thanks for the reply. EUT is very limited. Gigerenzer explains that human decision making is the best there is because all decisions have to be made with partial and incorrect information. I agree that machine decision making is not human and it is certainly not any better. I.e. Humans with EDS emotional deficiency syndrome CANNOT make decisions. We just use reason to gather information and then rationalize about the decision we made. This is by now quite commonly accepted by scietiests.

So your point is ... ( 1 month ago by ehsanul)

So your point is that we make our decisions based on emotion, and only afterwards create a rationalization of the decision made? While I'm sure that's true in many cases, it strikes me as extreme to say that all our decision making is emotionally based. A lot of decisions are based purely on reason, and these decisions would not fare so well if made on an emotional basis.
Can you provide a source that shows that people with EDS can't make decisions? I googled it, but couldn't find anything.

So your point is ... ( 1 month ago by ehsanul)

Thanks again. The ... ( 1 month ago by mjpucher)

Thanks again. The best read on the subject is Antonio Damasio 'Descartes Error'. He describes the medical history of people with brain damage to their emotional center. I have a friend whose son suffers from EDS. His son can talk clearly about the reasons why one or the other decision could be the right one, but when you ask him to decide he himself admits that he is purely guessing, as he simply does not know. Anyway, simply start to observe your own decision making next time in the cafeteria.

Thanks a lot for ... ( 1 month ago by ehsanul)

Thanks a lot for the video, I'll check it out soon.
So are you saying that your friend's son doesn't make decisions?
In the case of deciding what to have at the cafeteria, I can clearly see how emotion would take a part. But I'm thinking of a different type of decision, such as what type of algorithm I should use in my code to solve some particular problem. I'd just use whatever was best suited and most efficient, without any regard for emotion (it doesn't even come into the picture).

This may sound ... ( 1 month ago by mjpucher)

This may sound spiritual but it is a simple fact of life: As long as we see emotion as something negative that we have no control over then we resist this idea. Emotion is however the only source of power we have. Reason is empty and shallow. We never know all the details to be reasonable nor can we in any way judge what the result of our reasonable decision will be. So perfect reason leads to uncertainty or false decisions. When we feel one or the other way then decisions come easy.

Computers may self ... ( 3 weeks ago by 8thinpigs)

Computers may self improve but somebody had to make the computer.

But also... we use ... ( 3 weeks ago by ladonnaschild)

But also... we use the same energy in our earth and contain it and then give it all the information in the entire world. We have a fraction of it in our own brains and we are able to learn with it contained within us.. it would be insane to think that same energy cannot learn when wired up to networks and minds of all people in the world.

scary stuff ( 4 days ago by collapseofthedollar)

scary stuff

Tell a friend:

video.BinBrain.com

Entertain your eyes

Self-Improving Artificial Intelligence

URL
Embed Code