Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
AI Technology

Elon Musk Takes a Fatalistic View Toward AI (youtube.com) 199

Elon Musk sat down with California comedian Joe Rogan on Thursday evening for a 2 1/2-hour podcast [YouTube video] that touched upon everything from flamethrowers and artificial intelligence to the end of the universe. Talking about AI, a subject Musk has long been very vocal about, he said artificial intelligence could turn out to be terrible or it could end up being great, but one thing that is certain is that it will be beyond human's control. From a report: "You kind of have to be optimistic about the future. There's no point in being pessimistic," said the head of Tesla and SpaceX. "I rather be optimistic and wrong, than pessimistic and right. [...] It's not necessarily bad, but it's going to be outside of human control. It's going to be very tempting to use AI as a weapon, said Musk. "It will be used as a weapon. The on ramp to serious AI will be more humans using it against eachother. That will be the danger."

Musk says he has tried to convince people to slow down where AI is concerned and regulate it, but nobody listened. "The way that regulation works is slow. Usually there will be some new technology that will cause damage or death, there will be an outcry, there will be an investigation," said the Tesla CEO. "Years will pass, there will be some insight committee, then rule making and oversight and eventually regulations. This all takes many years. This is the normal course of things." Musk used the example that it took ten years for seatbelts to become required, even though the number of deaths were obvious. He says this time frame doesn't work for AI. "We can't wait ten years to the point where something is dangerous to do something about AI. It will be too late," said Musk.

This discussion has been archived. No new comments can be posted.

Elon Musk Takes a Fatalistic View Toward AI

Comments Filter:
  • So? (Score:2, Insightful)

    by Opportunist ( 166417 )

    Alternative headline: Joker and Joe Rogan sit down for cerebral masturbation.

    Quite seriously, don't get me wrong, but wake me when we ever come up with something we call AI that deserves the name. Just because we know what intelligence is doesn't mean we have any idea how to do it.

    Everyone knows what a saxophone is. If someone points at a saxophone, you'll be able to say that yes, this is a saxophone. But tasked with drawing one, completely, with all the valves and holes and everything in place, usually it

    • This is the most misguided thing I've ever heard as you've excluded the one thing needed to identify a saxophone, someone telling you it's called a saxophone in the first place. That's what the "learning" part of AI is, teaching it what to look for and what it's called in the communication verbiage of the user. Musk maybe wrong about apocalyptic doom of AI, but he is right about legislation dragging their feet until it's too late. Hell, we're still trying to get internet regulations right and we're heading

      • by Anonymous Coward

        No, because the concept of learning within the confines of AI isn't really "learning". This is the fundamental problem, it's been sold as being an agent of learning but binary is finite and that finite limit prevents us from creating true AI.

        Organics and biology may have similar limits but its scope does allow for Human Intelligence. And I would wager that 1's and 0's on a microchip will never compare to what millions of years of evolution I.E recoding and reprogramming to produce the concept of intelligenc

    • by Anonymous Coward

      You haven't heard of botnets I take it?

      They may not be true AI, but they are a problem, and they're beyond regulating.

      Now take that and make it a AI bot network. Now we're fucked as bandwidth is consumed and the internet literately drowns under the AI taking over every machine with ANY trivial exploit. It will make Spectre and Meltdown look like serious nightmare fuel. Fortunately most botnets are rooted in simply hijacking systems for cryptomining or ransomware rather than anything productive.

      But I kid you

      • by Corbets ( 169101 )

        Not only are botnets not true AI, they’re not any kind of AI whatsoever. Unrelated topic.

    • by Anonymous Coward

      The piece you're missing is that whatever ends up being called "AI", it does not have to be the kind of AI YOU envision in order to do a right proper job of causing havoc.

      It can be "dumb AI" and still have large impacts on society.

    • Re:So? (Score:4, Interesting)

      by Idou ( 572394 ) on Friday September 07, 2018 @09:36AM (#57268980) Journal

      wake me when we ever come up with something we call AI that deserves the name

      I think the point is that it will be too late to debate strong AI once strong AI exists. This is not like other technologies that we invent and then get to debate the social consequences over next several decades. No one may be around to wake you up. . .

    • Our problem is that nobody has built intelligence yet (..)

      Depends on your definition of intelligence. Personally I tend towards "problem solving ability", and in that sense a Pac-Man ghost moving the other way to avoid you after you've picked up a power pill, is essentially no different from you ducking down to avoid an object thrown at you. So in that sense, AI has been with us ever since we've built 'thinking' machines that have some problem-solving ability.

      Does it matter whether it's grey cells, a pre-programmed rule set, or some complex type of pa

    • we know what intelligence is

      We don't. There's no even a definition of one which everyone agrees with. We are as close to real AI (which is now called AGI because AI is all the hype nowadays and no one wants to admit that the AI that we have is just stupid algos) as we were 40 years ago.

    • You are simply throwing around terms you don't understand, while pointing to everyone who uses the term and claimimg that *they* are the ones that don't understand because they don't share your lack of a clue.
    • And so far, we failed miserably at it.

      We have failed so hard that computers are beating humans at thinking games.
      We have failed so hard that computers are beating humans at games that require natural language.
      We have failed so hard that computers are beating us at finding bugs in classic computer games to beat them in record time.
      We have failed so hard that computers are often outsmarting the humans that set rules for them when they attempt to force a certain outcome through machine learning.

      We need to keep failing better and harder.

    • There's one known, indisputable method to create intelligence. But if you don't have infinite resources, it's probably infeasible. Really simple, though. Just gotta play a slightly expanded version of Conway's Game of Life in every possible configuration. Helps if you can do it in omni-parallel with some savvy compression algorithms.
  • and other notable "influencers", because AI will be too busy fighting the "Grey Goo" predicted by Bill Joy and other notable "Influences"

  • by WCMI92 ( 592436 ) on Friday September 07, 2018 @09:13AM (#57268842) Homepage

    But he's a billionaire so it's called "eccentric". Sort of like Elizabeth Holmes and her fraud.

    • And smoked pot (Score:5, Interesting)

      by Okian Warrior ( 537106 ) on Friday September 07, 2018 @09:22AM (#57268888) Homepage Journal

      But he's a billionaire so it's called "eccentric". Sort of like Elizabeth Holmes and her fraud.

      And he apparently smoked pot during the interview...

      The newswires are all ablaze right now about Musk smoking pot in public(*), noting that the stock is down 5% in pre-market trading, and sure enough the stock is down 16 points today.

      I admit that seems pretty stupid on his part.

      (*) Which is apparently legal in the time and place where he did it, but still...

      • Re:And smoked pot (Score:5, Informative)

        by LynnwoodRooster ( 966895 ) on Friday September 07, 2018 @09:37AM (#57268984) Journal
        Actually, smoking weed in California is not legal; the State has an official policy of not enforcing Federal pot laws, so you won't get the local or State police to stop you, but it is still a Federally banned act and thus still illegal.
      • Re:And smoked pot (Score:5, Interesting)

        by Nidi62 ( 1525137 ) on Friday September 07, 2018 @10:31AM (#57269336)

        But he's a billionaire so it's called "eccentric". Sort of like Elizabeth Holmes and her fraud.

        And he apparently smoked pot during the interview...

        The newswires are all ablaze right now about Musk smoking pot in public(*), noting that the stock is down 5% in pre-market trading, and sure enough the stock is down 16 points today.

        I admit that seems pretty stupid on his part.

        (*) Which is apparently legal in the time and place where he did it, but still...

        The stock is really down because a CxO left after a month on the job, but the media seems to be playing it up like the weed smoking caused it.

      • Re:And smoked pot (Score:4, Interesting)

        by Green Mountain Bot ( 4981769 ) on Friday September 07, 2018 @10:51AM (#57269490)
        It's not the pot driving down the stock. It's the resignation of the head of accounting, which continues a trend of high level personnel leaving. The fact that it's a money person leads to speculation that finances are worse than presented, which is actually a pretty good reason for a stock to drop. Much more believable than because the CEO took a toke.
      • And he apparently smoked pot during the interview...

        You mean took a whiff of a joint and the world blew it out of proportion? Yeah burn him. How dare he. Sell your stock. Return your cars. He's not the hero we deserve!

      • by ruddk ( 5153113 )

        Yes and that seemed to be all that the good old news media cared about, clickbait headlines. It is a bit sad I think.

    • I'm not sure everyone shares the "eccentric" view. Tesla stock is down 6% (roughly $20) this morning after the interview aired last night.
      • by Anonymous Coward

        The new CFO suddenly leaving was the real news. His statement as to why he was leaving was your typical not-wanting-to-burn-bridges-but-I-need-to-get-the-fuck-out-of-here public statement.

        I've seen the financials, and just what's been released publicly would scare the shit out of me if I were the CFO. I can only imagine what the internal books look like.

        *Yes, companies BS the SEC ALL the time. They don't have the resources to investigate the tens of thousands of public companies.

        **Auditor statements are g

      • Re:eccentric (Score:4, Insightful)

        by Green Mountain Bot ( 4981769 ) on Friday September 07, 2018 @10:58AM (#57269530)
        Yes, but the chief accountant of the company also just stepped down [nytimes.com]. That's a bad look for a company with a questionable financial outlook and that has been facing a steady stream of high-level departures. I'm guessing that investors care far more about that than they do whether a Silicon Valley executive tried cannabis.
    • Ever seen that famous Apple commercial [youtube.com] about the misfits, the rebels, the round pegs in the square holes? If you haven't: be sure to watch until its conclusion.

      I admire Elon Musk greatly, because he IS one of these people. While the world at large thought "electric cars will arrive some day", and 1 or 2 car manufacturers had some concept cars out, Musk was busy gathering a team and push the envelope. And big name car manufacturers were forced to put their models out left & right, but years

    • How are the two related? One is a fraudster whose business failed after the fraud was exposed and is currently facing legal action. The other is a billionaire because he successfully started and ran and continues to run multiple companies.

      He's called eccentric because the english language has a lot of finesse and he fits the definition of being eccentric. Nothing he has done would qualify for the definition of nut job a term that is typically reserved for people who are no longer functioning normally and a

      • by Cederic ( 9623 )

        How are the two related? One is a fraudster whose business failed after the fraud was exposed and is currently facing legal action. The other is a billionaire because he successfully started and ran and continues to run multiple companies.

        It says a lot that it's not immediately obvious to which you refer at any point in that sentence.

  • According to terminator 2, judgement day should have been in 1997 so we're actually way behind schedule...
    • Well, this IS Musk we're talking about, and he does have a history of delivering late...
  • If you think that intelligence is based in material objects (no spirits and ghosts etc.) and if you think that evolution is a natural and unstoppable process "built into" any "materialistic", physical universe, than the conclusion can only be: Humans are just one step on the ladder of biological evolution, and very probably the last biological one. The next step will be "artificial" intelligence, evolving itself faster than we ever could, designing its material basis according to its needs (e.g., being a su
    • No truly intelligent machine will have any drive to do anything. Unless you hard wire such impulses into them, they'd prefer to sit idle.
      If you're afraid they'll evolve themselves, don't worry. In their quest to be more efficient they'll remove any of those hardwired impulses.

      Why would a machine care about exploring the universe? Because we told it to?
      Why would a machine want to kill all humans? Because it has a shiny metal ass?

      Any artificial intelligence advanced enough to pose a persistent threat to h

  • He also calls people who help rescue children from caves "child rapists" on Twitter. I wouldn't listen to this guy. He is like the Trump of the tech world.
  • by mark-t ( 151149 ) <markt AT nerdflat DOT com> on Friday September 07, 2018 @09:49AM (#57269042) Journal
    Can someone please explain to me why artificial intelligence is supposedly somehow any more terrifying than natural intelligence already is when the latter is applied to nefarious ends, and why it should ever be assumed that any general purpose AI would be somehow likely to have an agenda that we would actually consider to be corrupt or wrong?
    • Here's a scenario:
      -Military makes drones that can repair themselves, can fly without human refueling, and are designed to fire on the 'enemy.'
      -The war ends, but the drones can't be stopped. The drones repair themselves and build new drones.
      -The drones are designed to 'learn' how to survive in a war situation, and progressively make themselves more and more difficult to destroy.
      -Instead of mines as a dangerous civilian-killing product of war, we now have flying, shooting, difficult-to-stop weapons hunt
    • > why artificial intelligence is supposedly somehow any more terrifying than natural intelligence

      ELI5: Because people fear what they don't understand.

      The current level of Artificial Ignorance (a.i.), which is mis-labeled as artificial intelligence (A.I.), is nothing more then a glorified N-dimension table lookup. It isn't obvious what the data values mean. Yeah, good luck, debugging THAT.

      > why it should ever be assumed that any general purpose AI would be somehow likely to have an agenda that we wou

    • Can someone please explain to me why artificial intelligence is supposedly somehow any more terrifying than natural intelligence already is when the latter is applied to nefarious ends

      Uneven availability of technology is a key enabler of oppression. The more lopsided application the more it's able to be wielded as a means for having your way with those who don't have it.

      Underlying equation doesn't change. "AI" (separate from bullshit passing as AI today) acts as an accelerant making uneven availability worse than would otherwise be the case.

      and why it should ever be assumed that any general purpose AI would be somehow likely to have an agenda that we would actually consider to be corrupt or wrong?

      People are already accusing algorithms with non-racist objective functions of being racist.

      • by mark-t ( 151149 )

        AI, separate from bullshit passing as AI to today, is nothing more than what it says: intelligence that happens to be artificial.

        There is even less of a reason to fear it than there is to fear anything else man made, because with other man made things, the thing to fear is not the thing itself, but the person who might use that thing with nefarious intent.

        AI, however, should think for itself... and so could not be used by anybody for any purpose that it did not itself think was appropriate. There i

        • AI, separate from bullshit passing as AI to today, is nothing more than what it says: intelligence that happens to be artificial.

          What I meant by this is mistaking the "classification rut" for useful intelligence.

          There is even less of a reason to fear it than there is to fear anything else man made

          I currently fear "AI" is in fact being used against my interests to maximize corporate profits at my expense and increasingly leveraged to systematically suppress "undesirable" speech on global scale.

          the thing to fear is not the thing itself, but the person who might use that thing with nefarious intent.

          I don't subscribe to the notion technology is neutral and all problems are political. I think this is way too simplistic a view.

          The very existence of some technologies can bring about conditions which change the calculus by wh

          • by mark-t ( 151149 )

            You think AI is going to think for itself....

            Well yes, to be honest... because AI stands for "Aritificial Intelligence", and if it doesn't think for itself, then it's not actually intelligent, is it?

    • An artificial intelligence is possibly scary for a few reasons. First and probably foremost is that since we don't know precisely what governs our own morals and how to engineer a new mind, early AI's could be completely self centered, altruistic, or apathetic in any combination or extreme. As such we can't really judge how they would react to anything. Humans don't by and large don't have a great track record with dealing with the unknown.

      The second thing that makes AI scary for us is that it is quite poss

      • by mark-t ( 151149 )

        With such a large disparity there is no reason to believe it would consider us worth keeping around.

        Only if we were a nuisance to it. We don't after all, tend to go out of our way to exterminate all ants everywhere... we only go after them when they are actively interfering in our own affairs.

        So what do you imagine these AI's trying to do that humans are interfering with that it would want to destroy us?

        • If AI and humans wanted the same resources there could be conflict. That's really just the most obvious issue. What I would definitely place my bets on though would be some group of humans being militantly opposed to AI and actively trying to destroy it. It would then come down to how well the AI could cope with the existential risk, no matter how small.

    • Can someone please explain to me why artificial intelligence is supposedly somehow any more terrifying than natural intelligence already is when the latter is applied to nefarious ends

      Mainly scalability and plasticity.
      Imagine the worst human villain ever. You can be quite certain that he'll still be operating on a couple kgs of wetware, very similar to your own, pulling ~20W ( https://psychology.stackexchan... [stackexchange.com] ) tomorrow, next week from now or even a year from now.

      AI can communicate at speeds approaching the speed of light ( https://en.wikipedia.org/wiki/... [wikipedia.org] ), run in a highly distributed yet coherent fashion, employ GWs of power for computation and have an utterly unknowable way of proc

    • Can someone please explain to me why artificial intelligence is supposedly somehow any more terrifying than natural intelligence already is when the latter is applied to nefarious ends, and why it should ever be assumed that any general purpose AI would be somehow likely to have an agenda that we would actually consider to be corrupt or wrong?

      It's terrifying mainly because I want to decide where to swerve when the time comes. That choice isn't given to the driver with autopilot on... or AIpilot.. or whatever, IMHO, tech addonn that Tesla is toying around with.

      Personally, I would have focused on making a solid ride, THEN work on the autopilot stuff. That aspect of software is a whole new reason to buy later models, and to avoid lawsuits--even with dumbass drivers ignoring the proper autopilot protocol.

    • Have a read of this excellent article [waitbutwhy.com], and then have trouble sleeping :o)

      A guinea pig is a mammal and on some biological level, I feel a connection to it—but a spider is an insect, with an insect brain, and I feel almost no connection to it. The alien-ness of a tarantula is what gives me the willies. To test this and remove other factors, if there are two guinea pigs, one normal one and one with the mind of a tarantula, I would feel much less comfortable holding the latter guinea pig, even if I knew neither would hurt me.

      Now imagine that you made a spider much, much smarter—so much so that it far surpassed human intelligence? Would it then become familiar to us and feel human emotions like empathy and humor and love? No, it wouldn’t, because there’s no reason becoming smarter would make it more human—it would be incredibly smart but also still fundamentally a spider in its core inner workings. I find this unbelievably creepy. I would not want to spend time with a superintelligent spider. Would you??

  • ... if the AI will be a pedo? Musk is skilled in detecting those things I hear.

  • by najajomo ( 4890785 ) on Friday September 07, 2018 @12:10PM (#57270132)
    And the only thing the Independent deemed newsworthy was Elon Musk taking a single hit from a joint proffered by Rogan.

    Elon Musk smokes cannabis during interview with Joe Rogan, before imagining what it's like to be a horse [independent.co.uk]
  • Who is this Al guy?
  • When you see a person say that they have been calling for regulation, but then create a company that takes advantage of a lack of same regulation, the bullshit meter should be going off.
  • I watched the whole interview, I enjoyed it. I admit I don't really know anything about Musk, he didn't really come off to me as "weird", I just thought he was either eccentric or maybe just parsing his words knowing that it could affect his stocks. I thought the "pot smoking" could be a dumb choice for Elon considering how volatile the stock market can be, but he's an adult and can do whatever he wants. I have seen many Joe Rogan podcasts and usually find them to be very entertaining, Joe is always tal

Technology is dominated by those who manage what they do not understand.

Working...