Geoffrey Hinton: “Probably machines will get smarter than people in almost everything”

[Music] you the field I work in is called neural networks and what we try and do is train simulated neural networks to perform tasks and my graduate students and I made big advances in how to get these networks to recognize speech so we made a big advance in 2009 and by 2012 that was in the Android and now all the good speech recognition systems using your own networks recognizing speech and then two other graduate students of mine developed use the same techniques to develop a much better way of recognizing objects and images and so those were probably the two main contributions to get neural networks to be much better at speech recognition and at object recognition and that then had a big effect on the field of neural networks people started to believe that these things really worked and then lots of people went into the field and now they really do work I think neural networks are going to be used in a very wide variety of applications things like self-driving cars they're already being used for machine translation so it'll be a big benefit to humanity when we can all understand each other and neural networks probably in the next ten years will translate at least as well as people one of the biggest applications of neural networks is going to be in medicine they're already good at interpreting medical images and they'll soon be better than doctors at interpreting most medical images that'll mean you can get interpretations quickly and accurately so for example if you have a patch on your skin that you think might be cancer pretty soon they'll be an app on your cell phone where you just point your cell phone at the patch of the skin and it tells you whether that's cancer or not and if so what kind of cancer that would be a very nice thing to have and there'll be lots of applications like that also neural networks will be very useful for many different businesses in just making them more efficient [Music] with any new technology there's always new jobs created and old jobs that disappear because they're better done by the technology typically the old jobs that disappear are tedious boring jobs and the new jobs are more interesting so for example when you have a big mechanical digger to dig a trench in the road that replaces a whole bunch of people digging a trench it's not a very interesting job and there's more interesting things that people could be doing or as an information processing example if you consider an automatic teller machine in a bank those replaced lots of jobs in the bank they made banks much more efficient and I didn't think anybody would think that we shouldn't have had automatic teller machines there are some jobs it's clear we'll be replaced for example self-driving cars will get rid of a lot of people in emergency rooms because a large part of what emergency rooms do is deal with car accidents and we won't need so many people to deal with those because there won't be nearly as many it's very hard to predict what jobs will be created and what will be lost I do think there's a danger that people will lose jobs and I think it's very important we have a political system that uses the benefits that come pre-installed some people talk as though there'll be some point some particular moment when machines get smarter than people I don't think it'll be like that at all machines will get better at some things a much later they'll get better at other things machines will be better than people quite soon as interpreting medical images for example a bit later on they may get better at translating foreign languages it'll be a long time before machines are good at things like writing poetry or novels or understanding subtle really subtle texts so I think it's going to happen at different stages I think is very hard to predict how long it will take and most experts think it'll be like 50 years before machines become smarter than people at most things I personally believe that probably machines will get smarter than people at almost everything but it'll take a long time my belief is that the brain is just a fantastically complicated machine and that anything networks of brain cells can do we can mimic with computers and so I can't see any reason why we can't get computers to do anything the brain can do including having emotions so I strongly believe the in the future we will be able to get simulated systems that have real emotions in terms of having romantic relationships with robots I think most people who prefer to have romantic relationships with people I think that's what we're designed for [Music] yes it worries me a lot it worries me as a scientist because he's just very annoying when scientists have very clearly demonstrated something and people refuse to believe it and it worries me because things like climate change denial likely to cause a lot of damage to the planet I think one thing that could be done about it is just continuing to try to educate people better another thing that can be done about it I think is understanding some of the motivations of the people involved in climate denial climate change denial so I think investigative journalism is a very important tool there so investigative journalists can see who's funding news outlets that I advocate the idea that the reason actually climate change or that it's not man-made and so I think journalists have a big role to play there my case I came from a family where there were many scientists and as I was growing up I got the feeling that you could either be a scientist or you could be a failure so I had strong family push to be a scientist luckily I was also very curious about how things worked so I think the best motivation for doing science is just curiosity and understanding how things work and I don't think you can be a really good scientist unless your primary driving force is that curiosity so if we want to help other people become good scientists we need to foster that curiosity so in schools we need to not get kids to do a lot of rote learning we need to take their intrinsic curiosity and develop that I've always been very interested in how the human brain works that's why I got into studying artificial neural networks I think there'll be huge benefits to having a better understanding of how the brain works both in things like education and in healthcare but particularly in understanding who we are and so I think the thing I'd most like to see is that we get a much better understanding of how the brain computes

Channel: BBVA Foundation

Views: 12,551

Rating: 4.8869257 out of 5

Keywords: BBVA Foundation, Geoffrey Hinton, artificial intelligence, Frontiers of Knowledge Awards, Information and Communication Technologies

Id: rkiWi4Pdmzc

Length: 7min 53sec (473 seconds)

Published: Mon Sep 11 2017