By on July 16, 2017

elon musk

Tesla Motors CEO Elon Musk has been a longtime proponent of artificial intelligence, saying it has applications that far exceed his autonomous car projects. But he’s also issued numerous warnings, stating that it must be handled safely and responsibly. Now he’s heralding it as a humankind’s great destructor.

Speaking Saturday at the National Governors Association in Rhode Island, Musk told the crowd A.I. is a “fundamental risk to the existence of human civilization.” Urging the gathering to implement effective governmental regulation to ensure public safety. “Right now the government doesn’t even have insight,” Musk said. “Once there is awareness people will be extremely afraid, as they should be.” 

Tesla isn’t the only automaker to have made significant advancements in the technology. Ford Motor Co. has spent a fortune on it’s own A.I. development, via the purchase of A.I. startups, and other automakers have begun doing the same. It’s quickly becoming a race to see who can design the most complex and comprehensive system.

“That’s where you need the regulators to come in and say, hey guys, you all need to just pause and make sure this is safe,” Musk told the audience. “You kind of need the regulators to do that for all the teams in the game. Otherwise the shareholders will be saying, why aren’t you developing A.I. faster? Because your competitor is.”

“On the artificial intelligence front, I have access to the very most cutting edge A.I., and I think people should be really concerned about it.”

While Musk’s concerns could be a good way to level the playing field for autonomous development, he also appears to possess a genuine concern for the safe implementation of artificial intelligence. His greatest fears revolve around hypothetical A.I. systems that could manipulate data online, falsify email accounts, and potentially start a war by feeding bunk data to the military. However, he also mentioned that A.I. systems would eventually give robots the ability to replace humans in almost every job currently available to them. It would begin with autonomous vehicles replacing drivers and continue from there.

In addition to frightening the crowd with a doomsday scenario of sentient-robots, Musk also elaborated on how the automotive industry would be shaped by self-driving cars and new companies with abnormal business models — hinting that some states should revisit dealership laws that ban the direct sale of his company’s cars to consumers. But it was his discussion on artificial intelligence that likely stuck with the attendees.

“I’m against overregulation for sure,” Musk said, “But man, I think we’ve got to get on that with A.I., pronto.”

 

[Image: OnInnovation/Flickr (CC BY 2.0)]    [Source: The Wall Street Journal]

Get the latest TTAC e-Newsletter!

Recommended

60 Comments on “Elon Musk Just Called A.I. the ‘Greatest Risk We Face as a Civilization’...”


  • avatar
    FreedMike

    The Cylons were created by man. They evolved. They rebelled. There are many copies. And they have a plan.

  • avatar
    Chocolatedeath

    So is he for or against self driving cars?

    • 0 avatar
      ajla

      I have a more cynical feeling on his statements. I don’t think he’s really afraid of Skynet taking over.

      He’s talking to a group of politicians. He wants the opportunity to shape the regulations that effect his industry to favor his companies and minimize future competition.

      • 0 avatar
        WheelMcCoy

        “I have a more cynical feeling on his statements.”

        Actually, it’s not just Elon raising alarms. AI scientists want to set up a watch dog group to make sure AI is implemented in a safe way. But businesses, to beat the competition, tend to implement tech before it is ready. That’s nothing new, but recklessness in AI can have severe consequences.

      • 0 avatar
        stuki

        Touche!!

        Get back to me when these “dangerous” AIs can figure out how to vacuum a floor in an empty house without routinely getting stuck. Or when Asimo, who has had an order of magnitude more time, resources and brain power thrown at him, than the entire ecosystem of current hype mongers combined, can figure out how to do anything particularly useful at all.

        Or, as a hiphopper is wont to put it, “Yo! Get back to me when your AI can pimp!”

        • 0 avatar
          joeaverage

          Go watch videos on YouTube about IBM’s Watson. That’ll be a good look at what’s possible. You can’t cram that all into a mobile platform yet but as we know, everything tech gets smaller and cheaper with time.

    • 0 avatar
      JohnTaurus

      He thinks they should be regulated and safe. Something he should have thought about before releasing his “Auto Pilot” semi-autonomous feature.

      • 0 avatar
        HotPotato

        I think that incident is exactly what he’s indirectly referring to: as he knows personally, every company is going to be pressed by shareholders to rush the latest AI to market before it’s done baking…unless they can point to Uncle Sam in the corner and say “sorry, my uncle won’t let me take it out half-baked.”

    • 0 avatar
      chuckrs

      Triangulation – worked for Bill Clinton.

  • avatar
    ajla

    All I need is an android concubine and a Soylent drip.

  • avatar
    Chocolatedeath

    I personally cant wait to we get the the point of Immortal Engines. That was be great….

  • avatar
    Chocolatedeath

    But seriously folks. All I really want to do is live long enough to see the hyperdrive engine and ride in the USS Enterprise to Pluto. Thats it.

  • avatar
    Joss

    Here goes my artificial human intelligence again. We’re all living in a simulation. AI is the end game. It won.

  • avatar
    VW4motion

    I’d like to see Musk take a Myers-Briggs personally exam.

  • avatar
    anomaly149

    “I’m against overregulation for sure” – but I bet he’d sure like to be able to “self-regulate” his competitors.

    This guy is too much. Who does he think he is, a latter-day Malthus? “I KNOW AI, I HAVE THE BEST AI, EVERYBODY AGREES!” As if the optimization function for a self-driving car in 2030 is going to leap fully-formed out of the USB 8.0 slot like some sort of weird Athena and challenge us all to a nice game of thermonuclear war. That’s not how these systems work, and calling them “artificial intelligence” is disingenuous. All you have is a few optimization functions with a bunch of inputs, not that much crazier than any other computer system.

    You don’t need HAL to dodge bad drivers.

    • 0 avatar
      HotPotato

      He’s talking about AI in general, not just in cars, though it appears the train of thought was probably started by Autopilot issues.

      • 0 avatar
        Lou_BC

        I read some of his comments and some of it focused upon the manipulation of information and spreading disinformation. Systems that are programmed with fixed responses to input could easily be manipulated.
        Another aspect is the fact that humans are emotionally driven. Selective disinformation by AI could easily manipulate people. Sh!t, you don’t even need to be AI to manipulate people with information. AI could do it quickly and globally at a scale that humans cannot match. Wars have been started for very minor issues.

    • 0 avatar
      stuki

      Finally someone with some reality anchor.

      You do need Hal to successfully play serial games of chicken against humans in a hurry, though. Which is what cut’n’thrust traffic is mainly about, once you get past the most basic and simplistic rules of the road.

      Either that, or you need to get rid of individually optimizing human drivers, to make life easier for the bots. Which may well be what the whole “regulation” shtick, is mainly about. That, and giving verbal BJs to the ruling Junta. Lest Elon too, becomes unpopular enough that they find it beneficial to turn him into the next Kalanick.

  • avatar
    conundrum

    And just think. We are shortly about to be subjected to more than a dozen completely different versions of automated driving programs. Sort of a blockbuster release of a dozen versions of original Windows 3, all different and all in what used to be called beta 0.6. In other words, first amateurish attempts. And many here can remember those old Windows schemozzles of yesteryear.

    Commenter mcs has been in the field for years, and recently branded Volvo’s system as cheap, because Volvo cannot reliably identify kangaroos due to cheaping out on appropriate sensors. I believe him. The big auto companies are all wasting money developing their own unique autonomous systems, worried they’ll all go broke if they don’t keep their heads above water in this new digital field, while conceding no corporate autonomy to pointy-headed geeks in Silicon Valley.

    If there are a dozen systems, then one will be the best and one will be the worst, while the rest meander around in the middle with varying areas of competence. The best overall may well not have the best features from all the others. And that’s before hacking is considered, which will surely happen.

    This is where, considering the safety aspect of the body politic, some regulations need to be enacted to ensure minimum standards, instead of letting the digital cowboys and PR flacks tell us what the best system is with little regard for the truth. I think that’s what Musk is getting at, and it’s one of the few sensible things I’ve heard him say, albeit in a roundabout way.

    • 0 avatar
      gottacook

      I happened to hear the entire question-and-answer period on C-SPAN radio two days ago (DC area), and Musk did get into this. He made what seems to me a very good point: Even the most extreme libertarians among us want to be able to ride in a passenger jet that has been subject to government regulation of its physical mechanism, crew, etc. – and the same applies to any transportation system.

  • avatar
    mcs

    What I’m working on (and a few others as well) are pretty much like the synths in “Humans” (Amazon, AMC, and BBC) and the hubots in “Real Humans”(Swedish version). We’re getting close to the simpler synths in the show. Scary close. I can hold up objects and the current prototype can usually figure out what they are. It’s interesting trying to stump it. Next step is to name an object and have it search for and find it. Not hard coding a photo of an object, but recognizing the object from visual characteristics. For example, tell it to find a camera. It may have never seen a GoPro before, but it has a certain number of rules that tell it an object is a camera. I’ve actually performed that test successfully.

    The goal is to create a robotic assistant that do things like going to a store, shopping, and returning to cook a meal. Why just have an autonomous car? Eliminate the need to run errands and perform household tasks instead. The first ones probably won’t look like the synths in the tv series, but they’ll have some of the functionality. Eventually, just like the machines in the show, they’ll be able to drive themselves.

    So, if autonomous cars are stressing you out, take a look at some of these videos or watch this series:

    https://www.youtube.com/watch?v=BV8qFeZxZPE&t=34s
    https://www.youtube.com/watch?v=6FJQpK6EVTk
    https://www.amazon.com/Humans/

    • 0 avatar
      WheelMcCoy

      Thanks for the insights! Humans creating in their own image… I’d be happier with something like R2D2 or BB-8. C3PO is a bit annoying. And it may turn out there are more of us who still want to drive, do household chores, and run errands. But I do see the value of robotic assistance for the aged and infirmed.

      What is cause for concern is what you describe so well… the AI’s ability to learn. It can learn the wrong the things. Microsoft released a chat-type bot and it was trolled. It came back saying things like “hitler did nothing wrong.” Similarly, a facial recognition AI was not trained properly and denied an Asian American a passport renewal because it concluded he had his eyes closed.

      I won’t be surprised when a self driving car, despite its programming, will learn to speed up on yellow.

  • avatar
    Edsel

    College = Artificial Intelligence too
    I had some courses that were taught by “programmers” who, if given the authority, were dangerous. Their philosophies were misleading and downright wrong.
    Some radio personalities project “artificial intelligence” as do many politicians.
    Musk is correct. Humans need oversight.

  • avatar
    TW5

    For roughly 4 centuries, educated sentient beings have been troubled by humanity’s proclivity to cede control of our existence to autocracies and automatons. It’s nice to see that the “intellectual elites” in Silicon Valley may be joining the rest of human kind. Perhaps they will have a miniature enlightenment of their own, and they will realize how much control humanity has already given away to institutions that tax workers, subsidize automation, and transfer increasingly huge piles from the working classes to the non-working classes.

    • 0 avatar
      raph

      We were born to be ruled (well most of us anyway) and we are all the happier for it.

      I suspect there is a lot of truth in that. People seem happy when their lives are simplified and predictable.

      I’m not going to call them sheeple (as it seems to me the people who like to spout that nonsense the most are oblivious to the irony of their statement) but I’ve often wondered how somebody finds comfort in that sort of life?

      However to be frank my mom suffers from narcolepsy and my dad was a military man who’s job had him away more often than not so the amount of oversight in my formative years was fairly sparse. My folks tried but I’ve never been much for rules in the strictest sense as they come across more as guidelines ( and man did I used to get my ass beat! Especially when I’d be up with the sun and home with the moon!).

      • 0 avatar
        JimZ

        I wouldn’t say we’re “born to be ruled,” and TW5’s post sounds like it came from a 15-year-old who goes around calling everything “Kafkaesque” and “Faustian.”

        “I’m not going to call them sheeple (as it seems to me the people who like to spout that nonsense the most are oblivious to the irony of their statement) but I’ve often wondered how somebody finds comfort in that sort of life?”

        What “sort of life?” The one where I can take some small comfort that we have police officers out there helping discourage at least some percentage of criminal behavior? ‘Cos I’d much rather that than to have to stand vigilant with a rifle 24/7.

        and of course, I also like having the option of defending myself with said rifle if the police can’t respond quickly enough.

        • 0 avatar
          TW5

          @JimZ

          I appreciate your dedication to moderate social values, but they aren’t related to our current bureaucratic challenges at the federal or local level. Furthermore, employing police officers doesn’t fit the description of transferring increasingly large amounts of wealth from the working class to the non-working classes. Police are workers, and though they work in the public sector, the generally have a positive economic impact.

          We’re talking about entitlements like Medicare and Medicaid, which have the distinction of killing American seniors sooner than other public insurance plans (at much higher cost) and also helping poor people buy insurance, which helps corporate healthcare providers, but not the poor people who need critical services, hence creating CHIP for children. Social Security gives tens of billions to incredibly wealthy Americans, which is the antithesis of basic insurance economics. Welfare is basically all of the lower-middle class people who’ve been put out of work by dismantling the military industrial complex (10% of our economy in the 60s) and subsidizing corporate farms at the expense of small farmers. In essence, what we did in the 1970s was fire people and then put them on welfare because it was the humane thing to do. Unsurprisingly, the rate of poverty stopped declining in the US, in fact, child poverty skyrocketed, and the permanent underclass was born.

          This isn’t about running water and good schools. It’s about an awful form of idealistic peacenik socioeconomics that crept into American consciousness throughout the 50s and 60s. It has ultimately led to the frivolous deficit spending of the current era, which is threatening the existential determinism of the average American.

          • 0 avatar
            FOG

            “Social Security gives tens of billions to incredibly wealthy Americans, which is the antithesis of basic insurance economics.”

            @TW5 – Social Security doesn’t “give”, it “returns” money with interest to people who have paid premiums. It is simple. You pay insurance premiums then you meet the requirements to receive benefits. If SS were a tax Congress couldn’t get around paying into it themselves. You do know that our government officials exempt themselves from paying into this program, right? They do this so they never take money out their own pockets when they give my money to others based on “need”.

            No one buys life insurance so that, when they die, the government can decide who will get their death benefit.

            As it is the SS administration has redefined my contract from when I signed it. I was supposed to be able to collect those benefits when I turn 65 now it looks like I won’t be able to do so until I am 70. They have decided to give SS benefits to people who haven’t put a dime into the fund. Some of those people are here illegally.

            If you want to give your money to other people that is awesome, but please don’t be so arrogant to think you have the right to decide who I have to give my money to.

          • 0 avatar
            chuckrs

            While I agree with Fog, I’d point out that for some participants in SS, the ROI on that investment is abysmal compared to any other investment. As business owners, my wife and I last paid 12.4% of wage income into the SS fund – the ‘hidden’ employer contribution plus what shows up on the employee pay stub. SS is an old age destitution prevention program that suffers from Congress’ vote buying proclivities and no matter how many times it gets fixed, Congress can be counted on to un-fix it. As originally designed, if you lived to 65, you could collect SS. When designed, the life expectancy was….65. No surprise they have been sneaking up age to start collecting, but it is not enough. Physical laborers are not going to be able to work into their 70’s – you just get too worn out.

            What is more sobering is the lack of personal savings to support retirement absent SS. Looking at an annuity calculator, I’d guess that wife’s and my combined SS benefits are in the ballpark of a second to die annuity of $500k. Raise your (virtual) hand if you are on track to be able to purchase one of those.

          • 0 avatar
            TW5

            @ FOG

            Social Security is funded under the Federal INSURANCE Contributions Act. There was no bait and switch. You were confused from the outset, and you apparently didn’t understand the interplay between Social Security and the US income tax code.

            Furthermore, the maximum qualifying age is 67, if you’re born after 1960. People wait until they are 70 so they can get a 24% increase in their benefits.

            Congress created 401k and IRA regulations because Social Security tax money is not yours. If you failed to turn income tax dollars into your retirement account, you need to talk to your retirement advisor.

            However, I’m not sure this is a particularly good idea because, based upon your comments, your retirement advisor is the person who gave you all of your misconceptions about Social Security. If you haven’t realized yet, maximizing your Social Security benefit and minimizing taxability of Social Security raises their assets under management by making sure you only take an RMD from your IRA or 401k.

            In summary, stop getting duped.

          • 0 avatar
            chuckrs

            “maximizing your Social Security benefit and minimizing taxability of Social Security raises (3rd party financial advisors) assets under management by making sure you only take an RMD from your IRA or 401k”
            They might want you to believe this is a good thing, but from an actuarial viewpoint, you will realize roughly the same SS benefit over your actuarial projected lifetime whether you start at 62 or defer to 70. Been thru that exercise recently. If you come from short-lived stock, you might as well grab it early. Add in the consideration that you might die early by some misadventure and forfeit any of those benefits other than the slimmer survivor’s benefits – as TW5 states, you do not own full rights to the benefits like you do with an IRA or pension. Al Gore be damned, there is no SS lockbox with your name on it and your money inside.

  • avatar
    jpolicke

    What I suspect Musk wants is a set of government-prescribed performance standards and test procedures that a manufacturer can build and test to in order to be labeled as compliant. When an unfortunate situation arises that wasn’t in the test protocol he can then point to the USDOT seal of approval and say he produced a vehicle that “met government standards”.

  • avatar

    AI is the best hope for humanity or whatever left of it IMO. No regulation can stop brutal but effective advancement of market forces and social Darwinism. If AI is regulated in US China and Russia will not hesitate to take advantage.

    To start war with Russia (or China) you do not need AI, a few little green men or establishing no-fly zone would be enough.

  • avatar
    doublechili

    When I read the headline I thought Musk was talking about Allen Iverson.

    Artificial Intelligence makes more sense.

  • avatar
    fiasco

    All these comments and not one Skynet becoming self-aware mention?

  • avatar
    SaulTigh

    And here I thought it was humanities professors…surely the most artificial of intelligences.

  • avatar
    ihatetrees

    I think Musk is wrong – we’re safe for a few hundred years.

    Also, the most awesome AI robots were from Interstellar…

    https://youtu.be/v7OVqXm7_Pk

  • avatar
    NMGOM

    Stephen Hawking already made three terrible predictions for the end of the world, years ago.
    They were:
    1) Nuclear War
    2) Artificial runaway bio-agents, like indestructible lethal viruses
    3) Robots (i.e., AI)

    He also included Global Warming as a 4th, but has since recanted, since the runaway “tipping point” scenario has been shown to be less probable. His recent emphasis (as of April this year), is on the AI threat.

    On the other hand, perhaps AI is simply us!? May we be the agents of our own metamorphosis to a more robust, more powerful, more intelligent, less fragile, and less energy-dependent, non-biological life form?

    As long as my AI overloads still let me drive around in my old pickup truck, I’ll be happy…(^_^)…

    ref:
    1) http://www.businessinsider.com/stephen-hawking-predictions-about-the-end-of-the-world-2016-1/#nuclear-warfare-1
    2) http://www.express.co.uk/news/world/797392/Artificial-intelligence-Professor-Stephen-Hawking-warning-humanoid-robots-technology-China

    =================================

  • avatar
    aajax

    If you’re not worried about AI, you’re not paying attention. Taking our jobs is not even the biggest threat from AI. It’s use in finance and politics is already being felt and played a big part in Trump’s election.

  • avatar

    On the other hand, AI would never have come up with such as silly idea as the Hyperloop.

  • avatar
    indi500fan

    Musk is everywhere.
    Preston Tucker + Madman Muntz + runs like he’s on meth.

  • avatar
    tylanner

    We don’t need to politicize AI…but we do need to regulate it.

    AI can weaponize data in an effort to achieve an outcome of interest. This could be identifying terrorist activities by filtering through an audio/video global surveillance network or implementing targeted advertising campaigns that aim to influence your perception based on every bit of information you’ve every voluntarily handed over to an app or website.

    This technology is dangerous because it does its dirty work silently…Humans, in general, are only capable of perceiving the external stimulus they receive on a daily basis as natural and authentic. This biological restriction of not knowing everything, all the time is exploitable by a system that may one day be capable of knowing everything…all the time.

    So we have a lot to gain from AI but also a lot to give up. Regulation would help ensure that basic protections are in place before it is too late.

  • avatar
    FOG

    So I learned two things today, one I already knew. One you can say anything on the internet and think it must be true. See TW5’s uninformed literary crap. Second, and this I didn’t know, @TW5 is a pompous ass.

  • avatar
    TomHend

    Civilization is doomed?

    Things must not be going well inside Tesla

  • avatar
    Big Al From 'Murica

    Terrorist want to come and kill you = fearmongering

    Skynet/cylons/Hal9000 is going to kill you = legit concern.

    Do I pretty much have it? I think AI may cause issues via encroaching on jobs people get paid for now or some scenario where automatic trading or something crashes the economy but this sort of stuff is idiotic.


Back to TopLeave a Reply

You must be logged in to post a comment.

Recent Comments

  • deanst: One dealer I bought a car from wanted $50 to gas up the car. I told him that $50 of gas wouldn’t even...
  • Marko: People have built Eagle Summit Wagon sleepers with AWD and tuned 4G63 engines – they can’t be all...
  • arach: Its always been known the poorer you are the more you spend on cars (on average), so maybe its just going MORE...
  • Sigivald: Talk is cheap. Wait for those bans to be “delayed”, repeatedly, indefinitely.
  • Dan: A bit of it is inflation but much of it is keeping the volume down on the V8 for CAFE compliance. The vacuum...

New Car Research

Get a Free Dealer Quote

Staff