WWL>Topics>>8-21 10am Garland, Superintelligent machines

8-21 10am Garland, Superintelligent machines

Aug 21, 2014|

An Oxford professor of philosophy says the greatest danger to mankind is “superintelligent” machines…just like we saw in the Terminator movies. Intelligent machines could see the human race as a threat. Do you think that is really possible? Is technology advancing so fast that we cannot possibly keep up? Garland is joined by the Professor and the writer of the article, Professor Nick Bostrom, Director of Future of Humanity in the Oxford Martin School at Oxford University and Dylan Matthews, writer with Vox media.

Related Audio:

  1. Think Tank 1210pm drug addiction in the city


    Tue, 28 Mar 2017

    Should drug addiction in the city be treated as a health issue or drug issue? More deaths due to overdose in New Orleans than homicide. This hours guest: Dr. Jeffery Rouse - Orleans Parish Coronor

  2. Think Tank 1110am healthcare plan


    Tue, 28 Mar 2017

    Bernie Sanders said he’s going to push his plan for a single-payer healthcare plan like Europe.  He says Obamacare is costing us too much and the GOP can’t get their bill together to correct the problems. This hours guest: Michael Cannon - Director of Health Policy @ Cato Institute

  3. Think Tank 1010am recreational marijuana


    Tue, 28 Mar 2017

    OH CANADA!  Could Canada be the next country to legalize recreational marijuana? Canada is proposing legislation that would legalize recreational marijuana by 2018.  This hours guest: Chief Larry Kirk - Retired Chief ( Old Monroe Police Department, Missouri & member of LEAP (Law Enforcement Action Partnership)

  4. Think Tank 1210pm select committee


    Mon, 27 Mar 2017

    Is an independent “select committee” necessary in the investigation of Russian hacking & possible collusion with Trump associates? This hours guest: Max Bergmann - Senior Fellow at Center for American Progress Steve Bucci - Director of the Allison Center for Foreign Policy Studies at the Heritage Foundation


Automatically Generated Transcript (may not be 100% accurate)

Welcome back busy area of great interest today in the pretending. Drove known will have -- -- -- a -- in New Orleans is huge turning creating a new unit to ensure there are convictions. In this city. We'll get into details. About that and a lot of moved into -- is seeking its in the whole hour. Level -- caller go talk about the beheading of the American journals by ices and brought. And talk about some things that President Obama six did. And what can be and if anything. At this point and done. Be you know. But -- -- -- works news I tell you what I think and then. We set about showing where I am brawl. I called that the scientific -- not conservative. Liberal. Reduced to -- subject every little bit about it Terry what do think in -- we try to disassembled or blog believe so far. And I gotta tell him the last nine years it's. Eighty person that Tom the only person that time -- proved wrong. So what upstart doing. Reading a lot about science in particular would become soon. Super intelligent so artificial intelligence. Don't have to -- booms that are integrated. And there reduced but you -- website call box. And it's a personal opinion but I think it's one of the best I've ever seen. -- it gives you the combination of headlines and without talking about Britney Spears. And did also is develop a card system. On very complicated subjects or the most interest in -- -- Where in a very short time and relatively easy to understand. Card by card by card. Within a couple of minutes -- working foundation. On the debate to the issues in the news of the day of the year. And as usual is good to every details on balked at how I solve this headline. In his. This Oxford for poachers thinks artificial intelligence. Will destroy. -- all. And both. Our producer Todd and some line. Went through a lot of work to -- get. Artificial intelligence experts listed all over -- the country. And tugged at the at the beginning of the -- said. We call the professor and so that's proposed captured on. And he -- well. I'm gonna call him and the author of the so put those two. They'll they'll never come on the lower bowl gift was right once again I'm brawl -- professor nick bought from. Director of the future of humanity in the Oxford -- school and Oxford University professor thank you so much for joining. -- If you don't mind I'm started the top of this the report. And go right on down through this sport is we can give that time -- And you wore a professor at Oxford. Area. Of what loss of -- Yet hit and I am and like -- run this multi disciplinary research institute there. -- -- -- them mathematicians often persons scientists. -- trying to think carefully about big picture crest of argument that. It's interesting. Businesses on the short disability you're grown other road -- small book called week you know in time. And his ideas that were our computers and and society. Is. No -- burrow on the the computer time world of the law. That says the computer will double in power every twelve to eighteen months. He feels like we you're role and expect the central time now 2486. Team and deserved read your article that says artificial intelligence that at some point. Many experts believe that artificial intelligence. Would -- -- point we're not only exceeds human intelligence. But who's capable of extending its own intelligence. Setting off an expert eventual intelligent explosion. Job keyboard you mean by X prudential intelligent. Well if so whether it is exponential or not. The ideals and intelligence explosiveness that. Whereas now it's human beings -- -- computer science research. And developing the next generation -- -- At some point if they AI becomes Smart -- they will be the ones during the research. And from that point onwards whenever and I get a little bit smarter. Failed to get even better and improving themselves even products. -- There is the scenario in which it would have once you reach a sufficient level. Maybe slightly about human level. -- very rapid transition to radicals super intelligent. And back and would be an intolerance explosive so that's. A lot of the and that's an artist that -- -- in in in this book. Super intelligent but it's just coming out. In new mourned the paragraph is for real to Woodward -- -- is gonna do for human Indians and you've brought a couple. Super intelligent computers could resolve. Really talked scientific problems. And it could. Holy replace human workers and Barbara about -- suitable that everybody's fear but the news enabling humankind quit working. How how -- that -- Well some time at the moment machines. Can substitute for human in particular -- absolute. But you've got in factories. A lot of fault permissions. We got software that can sort of do some of the things secretary used to give. But we still have to develop -- of one new machine for each type of job. When machines. Require. The same general learning ability that that we humans that they single system could then learn to do and it. And all jobs. And substitute for humans not just in terms of physical labor. But also in terms of intellectual labor. And economically we have to re range everything prudent way because. Who would we tell. Who profit Lopes who gets the biggest sellers who we quote quit working. How would we determined to you -- upper middle and lower class as -- would all that -- hazard beneath all along those lines and. Well that has been some ball. And all of this says there is a lot of uncertainty. And at about these things happen but it is I think as important as the starting point to to realize that the development of human level machine products. It would not just beat yet one more cool technology. One more step forward -- community. I -- watershed moment in human history perhaps the most. Important things that has ever happened. I'm something that might determine the entire future of art pieces Satan when that moment arrives it will be -- kind of pivot point. Around which everything hinges. Some would say this is terminator stuff worth thinking about -- I don't think that gives it Kolb. The potential. Of this being. Real. News news news news is moving report. I'm talking about. The report on the box where apply it in the -- The talks -- professor. -- artificial intelligence and will destroy also ball. And his name propose certain that bought from. Director of the future of humanity and doctor Martins school at Oxford University. A professor won one of the the trip reports on this that I woes talk about it and was verified in the article about it renewed. So the one of the problems is if you convict -- super intelligent computer and and we have been or doing got more and more. But how do you read wisdom decreed -- Well so that they -- unsolved problem not mean basically back to problems as the one -- how to create and machines that aren't relevant in the first. And that's the problem that tens of thousands of people working on around the world the costs are big economic incentives to make progress. Then the other problem -- it occurred to create machines that aren't Colin Powell could you make sure that they would be safe and beneficial. It is to control problem -- control of super talent. And that's the problem where maybe half a dozen people around the world are working. And yet it is going to be essential that we have. -- solution to the second problem the control problem before. Somebody figures out the solution to the first problem how to build a guy. Do you think the scientific community sees that could you said. Hundreds of thousands of people around the world working all and machine intelligent but very -- working on control. Yeah and 21 of the things one of the reasons for writing this book. Now on the superpower that is to try to begin to change that I think that has been growing awareness just very recently in the last year to. That this is really neglected -- that that we need to begin to -- and I think that. Some of the brightest minds of our generation and the next generation should be focusing. Specifically on this problem figuring out how you could continue or super talents. So that it would be -- We don't now how long it will take. To get to that point of maybe it'll take. Ten years or fifty years or under here is nobody knows but by that time have to have cracked it's other its control problems. You know we're gonna blow -- efforts are reading about this couple years ago my thought was may be somewhere in the next to century. But to there's a poll that mentioned in this article. And the experts who were polled. Said human level machine intelligence. Will be -- spied for reporting your 25 to. While not a that but what if the person's confidence. All of that and so I think that's yet as the future uncertainty around that. And and it's important that time. That we start case. Technology it's uncertainties that rather than either ignore the prospect because returned safe secure land. Or pick some arbitrary date that we need to think about the distribution. Of possible rivals -- the end and kind of plan accordingly. -- Do you being. That that rather than more shall all with computers -- or moving. Exploded and truly you and and the very definition of expert -- true wouldn't -- -- -- That what seems like broad group will be column a lot of progress in the in huge amounts of progress and then. Progress that we may not be able to control. Yeah I think it. -- the hypothesis. That -- like one they've been intelligence -- can be distinguished from any claims. About -- exponential curves. And perhaps it will not be exponential maybe there will be a period of stagnation for awhile and then maybe at progress -- -- again. I think however that maybe once you -- The component -- calls to human level general incompetence. There are these strong feedback we've spent. I think likely to catapult us from there on cricket is super intelligent. I'll tell her sudden this. I did I did read Hubert Dreyfus whose arguments. And and basically he disagrees review board what you've -- with the others sought to ensure the other. Philosophy professors in take a look at this and disagree. That would depend on precisely. Why it is stated to Greece and its different people might have. Different object. Like -- says dude you can model things in this way. Yeah I mean I think it is basically wrong well it's if it's true that we can do this today like -- -- nobody. Claiming that we have. Human level AI today I don't think it's very likely that it's gonna happen in the next two years or even decade. Decrepit whether we will eventually reached this point. Perhaps would have liked -- a lot of you listeners out there this. Should be taken seriously. -- we haven't existence proof -- the human brain. It's small physical systems that somehow produce its general incompetence. If we don't first figure it out as some other way we will figure it -- by gradually understanding more and more about how the human brain works. And -- of one of the sentences mentions. Literally copping a particular human brain. Is that possible. What it's not something we can do today but it's one of -- several halfway towards. Human level machine intelligence most of ai is based on. -- up abstract mathematics and not contribute -- to mimic what the human brain -- but the different approach. Is to try to reverse engineer the human brain. Learn from biology and do what nature has done in computer and -- deliberate thing. -- but apologies to develop technology to. To span and and upload a particular human brains in the way but would work. Would be that he would. I'm not freeze that human brains like it opted to thin slices and then he'd go through. Operates a microscope. Think Tikrit the backup images that you could then use -- to make -- image recognition. Algorithms to extract the three dimensional -- network. And and emulate that in computer that's something that we're very far from being able to do today but it would not require any deep theoretical. Breakthrough to do it we get faster car a lot better. Technologies scanning technology. Image recognition technology. So is the basic concerned. He's the terminator. The taking over of computers in the far distant future. Or. Simply putting out rhythms in London and and computer computations. And getting something out -- we don't know how we guarded or we didn't plan and get. Well both -- in -- and super intelligent I make the point that it's not. The bodies of of the future a guy that we need to be afraid opted not -- robotics with machine guns and rockets and 88 the brains. A very high level of intelligence is an extremely powerful thing he can be used to develop new technologies quickly to make plans. And that super intelligent -- even if initially. It doesn't have any arms or legs. It could use its intelligence. And for example to manipulate humans to gazans human accessories that it continues to get access to physical manipulators. It's important not to have to put more five and back and say -- we -- -- a lot of our thinking is kind of shaped by what we've seen Hollywood movies. Have but adult content scenarios are extremely possible that more possible kinds of Mars are a lot more boring -- And because there are harder to describing in concrete detail. It is and it is part of it. And -- gonna have the strongest part of -- having the computers say very -- in the future with super intelligence. That convinced the coincidences via its screen convinces by its ability to. Debate or sell an idea. To convince a human with the lay eggs to carry out. Whatever they like to see done is that an exaggeration. Not that that would be one way it's not. So the 11 idea that comes up is how could -- keep it separate elements safe but the idea is wildly cupid in the box you lock on applauded the Internet cable. And it may be only communicate. Weighted by allowing it to that answer questions on the screen by typing text. That's how could it possibly posted dangers if it didn't have -- arm's sore legs. About we know of achievements are not secure system to mean even. Even human actors often managed to find gullible humans and couldn't persuade him to do their bidding. They put out a super intelligence -- as superhuman level of preparation. The only safe assumption is that it would cricket under way to pocket play out of its box. And the end result would probably be the same I think we had given it access the Kia typical manipulators have from the outside. Professor of before we lose YouTube would give the audience then title your book out of they've -- Super intelligent -- dangerous strategy. It available non. Now on the book in almost on and it's coming out in force here in the US on the third of September. But I'd love to be W Gupta back on the show when we got more time thank you so much for calling it an -- of Simon and called me and says. The professor. He thinks artificial intelligence. Will destroy us all. And go to -- incidents and like moss and audio skipped Watergate and Fukushima. Get some noted that. Put attempt to look at the professor books super intelligent dangers. And strategy used its nick bought from. And we're just had a Mullen is the director of the future of humanity. And the Oxford -- school at Oxford University. And it's it's of fascinating. Interview of you would read negative tone in new website. Call box do you OX. And the headline that -- soaring as -- it is the Oxford propose different things artificial intelligence will destroy civil. And he ended start tilting at some point. Mixers relieved that artificial intelligence. Gonna branch two point. Word inaugural week's siege human intelligence but it's capable of expanding its own intelligence. Setting Altman. Excluded too intelligent. Explosion. And they drew like I was a year ago and I don't know whether the neutral laws and we're done any number of stories on film that was too -- and -- war in -- eighties meant 1632. Via physics professor have. Earlier. Water cup -- -- Had given me an example about a -- of what we just got a hole. Of the author of this or removal to get a few minute with the and so -- right to him Dylan Matthews I appreciated that time thank you for calling -- Or just filling your audience. I did twenty years in the news media business and as senator from Simon and -- call me and said hey. That professor of things artificially intelligent. And destroy -- over all. My response would have been their but he posted Watergate and -- machine and I get the reason for culture. -- what what were your thoughts after you talk to the professor we're used skeptical. Going in the and did you change at all either way -- come and -- When you wrote to what -- -- attitudes towards this possibility. And what was it when you were finished researcher. Developing stroke Edward. He's. Appreciate it is groups. Tikrit here. I was skeptical going in the book coming out I think in general people and you underestimate how hard big. Technical. Accomplishment like artificial intelligence so -- Home. I you have a hard time -- But the weather's going to do Clinton change you can break her so much more complicated weather systems. Did it in the Kyoto skeptical that we're gonna get there's. That he thinks there. That the I appreciated. The argument that you took that is the kind of that I might have been a little skeptical about -- boy what an article about -- -- actors sound card and four. Is what consequences that would be a threat. I was impressed by early in mapped out to different possibilities then why. If that happens. -- -- to work it. Well I was like your program more so I just totally discount it because. Birds implicitly it's a couple of top box and scope wires. It's leagues and norms don't work to world at that point in time. Baldwin he would explain. Something was super intelligence working with the people who work with -- If you duo computer might be able to convince you -- in some former fashion. To become an agent for the change he wants and -- it's super intelligent you -- even realized. The -- you're doing is working towards his goal. Did that. Did you buy that at all. I agree with that and that heat lately I think that perfectly clear that certain about the company. Including what terminator where we're you have bots that are super intelligent look like humans that are I think. -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- And instead. The hardest thing I think getting in our official talked to people around. -- understanding certain emotions. And nuances and and persuading -- in particular. -- inclined to do is exceptionally difficult technical shipment. Far greater emotional intelligence sentence and machine that equity -- And so. I guess they get backed you management team skepticism about how we ever -- there that I think. Create a machine that is so Smart and so perceptive about people. That is able to persuade people to -- things. For it that it would otherwise do. Don't really accomplish them back in time and time I think -- Convinced then press preposterous that workers here. I agree with the but he also called my attention would be. Ordered out hackers. Are bad they get into all or send says. And and a -- click here from get a million dollars and more drew really doing your opening your light spoke to him. Something. That you had no intention of doing and certainly. Weren't even worth the abrupt with coach and so thought that was pretty good correlation tell me this. The professor and mentions in his book his department. Go to Poland may ask the question Bob what you're do you think Thursday 50% probability. We'll have human level machine intelligence and the media in the averages enter toward reporting your 25 two did you look into the -- lower. Take a look at the way they did. And I did. I think. And -- about the current and came under I think welcome back to hold the researchers. I think. -- dear friend and dance and certain users -- -- The media institute figures. And it brings a lot. More to the point that concern -- -- thoughtful and open things in the peace it is. If you're someone you've gotten things that there's going to be a lot of progress and our national intelligence. And it doesn't a couple of sentences you to become an artificial intelligence picture. So that people in this poll are people who are that is extremely promising -- -- producers. I'm in doing it out of their corrupt or anything it's a bit. It's flat for people who're we're going to be more optimistic about it. -- and smarter people people and arbor area. I'm gonna -- Turkey certain bias to work out I don't want to buy anything serious about the Rutgers there are people who. Apparently important topic. Bit but it doesn't give you. Having too much faith in him. Prediction like that. Very accusing Julia appreciate the time killer body bucks keep doing what they're doing and I think a problem getting back to some journalism from the floor at the web -- Tony don't change. -- -- -- -- -- -- -- -- -- Well we're talking about PO box web site. It's in usual there's at least it is can be. They'll take 45. Or or ten. Pretty difficult subjects cords would forge and -- Ferguson. And they created card there can be tend to have been Bob them water over. And as you flip through them you learn. The most important facts about that issue. And a very short -- simple. Easy to understand format. And by the time you get to 25020. You've got a pretty good working. Idea. As to what's going on. And threw for a Wikipedia. Did and talk to the people who in the profession and I don't see so far. In any conservative or liberal by house basing that they're trying to do something. It's unique can bring you were never -- pork thought up. Gathering information and assume and double dubbed.