Elon Musk: ‘With artificial intelligence we are summoning the demon’

“Tesla chief executive Elon Musk has warned about artificial intelligence before, tweeting that it could be more dangerous than nuclear weapons,” Matt McFarland reports for The Washington Post.

I think we should be very careful about artificial intelligence. If I were to guess like what our biggest existential threat is, it’s probably that. So we need to be very careful with the artificial intelligence. Increasingly scientists think there should be some regulatory oversight maybe at the national and international level, just to make sure that we don’t do something very foolish. With artificial intelligence we are summoning the demon. In all those stories where there’s the guy with the pentagram and the holy water, it’s like yeah he’s sure he can control the demon. Didn’t work out. — Elon Musk, MIT Aeronautics and Astronautics department’s Centennial Symposium, October 24 2014

Full article here.

106 Comments

      1. Samaritan. Not as well known as the others, but at least it has an ongoing continuing story.

        If we do create God (not exactly in our image) we’ll get just what we deserve. Assuming, of course, that we haven’t already created it. It might just be biding its time, hiding out in the net…

    1. I can see his argument if prefaced by the following ideas (your opinions may vary).

      In the US Education quality and results have plummeted to all time lows. I’m in my 50’s and see that kids today graduate highschool with fewer skills and abilities that I had in 8th grade in 1975. The government, both parties (are actually just different sides of the same coin, and only serve to enable and perpetuate each other) is making that worse and shows no signs of improvement. Each year we release fewer people that are ready for life. A 2 year college degree doesn’t even bring them up to what a 1980 highschool graduate was and most education professionals will candidly admit that the first 2 years of college are remedial. In short we are producing 95% drones.

      In California, where I have the dubious belles sing of living, we have what is almost the worst public school ratings in the nation, with some of the most sought after universities that put out the top 1% of thinkers. That too is declining in quality though. Yes we have more information and data, especially in this information age of the WWW, but Information is not knowledge. It’s the ability to apply that information to get things done is the power of knowledge. What is being produced in the universities are a batch of highly informed parrots with little ability to apply that information broadly. These best and brightest though could not do very little of value outside their niche.

      Most of these also have a vary narrow idea of what life is for anyone else. Yet these will the architects of what will become AI. That AI will be based on their world views and values. That will set the expectations of acceptable behavior the AI expects, but will not encompass the average man & woman, let alone differing opinions or constructive descent.

      Eventually as the intellectual foundation of society decays further and the increasingly fewer top students become too few to support the systems in place, we will turn to AI to take up the slack on simpler decisions and tasks. After decaying more and seeing the AI can now do what we can no longer do we will ask the few intelligent ones to enable AI to do more, but ale by now those sake to add this extra ability to the AI will not be up to the task and will not even know what secondary and downstream implications there will be. They will believe they are just making another tool to do similar to what was done before (with just a little tweak) but once activated they will not have control or Understanding to see what is now happening.

      All this is not only plausible most of it is now fact and pointing toward the overlord scenario.
      With all this coming I wish I had not brought children into this world.

      1. Hey old guy, high school graduates weren’t rocket scientists back in your day, however you happen to remember it. Most people are sleepwalking. That was true 2,000 years ago, 50 years ago, and today. It was one of Socrates biggest disappointments. So your condemnation of modern education falls a bit flat. Regarding AI, if anything with true intelligence is actually developed (and not mere mimicry of intelligence) it will be developed by someone with deep insight into Being and Truth, by methods not obvious to anyone else. Will the AI be benevolent? Surprisingly, yes. And lonely. And creative. So fear not. (It’s at least 50 years away, anyhow, btw.)

          1. The last moonwalk was 1972, so the rocket scientists were from an earlier generation. But rocket scientists have done pretty well since.
            Most lamentation of declining standards is nostalgia, and clearly discounts future opportunity. Life is much better now than 1975, and will be even better 40 years from now.

      1. Exactly, excellent insight on your part. Machine intelligence, so called, theoretically should be perfect, benign, beneficent. But AI is a human construct developed according to human goals and objectives, so it is ethically and morally constrained, and thus, conceivably, insane—as humanity itself has convincingly demonstrated itself to be, with its self-destructive behaviour and its tenacious clinging to absurd beliefs. But maybe it’s only me that realises that.

        1. would clinging onto me tenaciously be an absurd belief?
          Especially if you believe that squeezing my juice and fermenting it into cider vinegar then mixing it with honey is good for your health?
          Squeezing my juice and fermenting it into cider vinegar is good for salad dressing? (ps. I love dressing up Salad, she is well beautiful in all the right places!)
          Squeezing my juice and fermenting it into cider vinegar, you can use my fermented juice for cleaning, baking and pickling? 🙂

          1. I like the way you think. This ones-and-zeroes business is decidedly recherché. Wetware is the future; — not to mention the present, and the past. Futurist Isaac Asimov was one who understood that robots would come to be shaped and conditioned like us, and resemble us, and become our undying companions — companions who would nonetheless mourn our mortal passing, and decommission themselves in their grief.

        2. Developed toward human goals and objectives. Great, let’s look at that a bit deeper.

          Are we talking politically correct goals, robber baron capitalist goals, terrorist goals — etc.?

          I fear the programmer.

        3. I’m with you sister! We need to seriously grow up as a species. The kiddy game playing has got to stop. We have some serious survival challenges that require adult rational, and emotional, thinking.

          But game playing is important for our sanity as well.

        1. http://en.wikipedia.org/wiki/Eliezer_Yudkowsky

          Yudkowsky’s interests focus on Artificial Intelligence theory for self-awareness, self-modification, and recursive self-improvement, and on artificial-intelligence architectures and decision theories for stable motivational structures (Friendly AI and Coherent Extrapolated Volition in particular).

          Need I point out: Incredibly bad idea. Limit AI to purposes and tame it to be a tool for mankind, never anything approaching ‘self-aware’. That’s just lunacy for coders who want to pretend they can be GAWD. Total silliness. (IMHO of course).

    1. If by “our human ones”, you mean the US, I’m guessing you haven’t lived in Somalia, India, North Korea, etc. etc. etc.

      But if you are including those, I’d suggest that bad as it is, it could still be much, much worse.

  1. Musk worked hard at sounding profound. Except he’s a dork.

    He came up just short of pronouncing (queue deep male voice with heavy reverb) “I want to protect humanity from itself, for humanity is still in its infancy.”

    Hey, Musk: Answer your phone. Your mother ship is calling.

    1. So you think you are qualified to school Elon Musk. You think your comment is more profound than his. Elon is more of a dork than you. Hmm.

      The problem of people being unable to provide a worthwhile opinion is exacerbated by the tendency of those same people to have no idea that they lack minimal expertise. Its a tough problem.

      1. Put in more simple terms, 82% of American drivers think they have above average driving skills. True dumb doesn’t have the skills to recognize itself or the grace to defer to its betters. (He said, assuming himself to be a member of the latter group.)

        “I’m sorry Dave, I’m afraid I can’t do that.” wasn’t a joke.

        1. “True Dumb” — wow, I have never heard that before and it would make a really catchy reality series title, better hurry and copyright it. Of all the varieties of dumbness, educated dumbness is the most interesting and fun to watch, for the pop of the balloon is loudest with an inflated ego. – been a victim of that myself. 🙂

        2. Food for thought:
          One of my favourite sayings is –
          “50% of the people out there graduated in the bottom half of their class”. Unfortunately this applies to everyone, tradesmen,doctors, plumbers, teachers, clergy, and so on. Another problem is that average class performance level is dropping steadily, year by year.

          Another one:
          “If you have a 50-50 chance of getting it correct, you have a 90% chance of getting it wrong.” (especially if you are a politician!)

          Cheers!

          Bob

      2. Musk’s opinion speaks volumes about him. As for me, I’m a mechanical engineer with 16 patents. You, are probably under the age of 25 and are too stupid to know how stupid you are.

        1. Musk’s opinion isn’t new. Think back a few years. Ted Kaczinski said the exact same thing. Only with, shall we say, a bit of extra emphasis. Personally, I think they’re both right.

        2. Funny, but your gibberish is just as unintellectual as the tripe that lazy slob video gamers type in-between bags of Doritos.

          The difference is, they know how stupid they are. You seem to think you’re smarter than the rest of the planet — and thus you deceive yourself.

          1. Smarter than the rest of the planet? Naw. Just clearly smarter than you. And, since I was designing adaptive feedback algorithms for 8051 microcontrollers before you were probably born. So—and just pardon me all over the place—but I don’t place much credence in your opinion on any of this.

  2. The problem with all this brouhaha is that we have been “10 years away” from good artificial intelligence since about 1955. The truth is that they have not yet really demonstrated anything that could be even slightly considered “artificial intelligence” in the sense of a self-aware program, and there is absolutely no reason to think that they ever will. We will probably someday be able to create a program that will mimic human intelligence, but I seriously doubt that it will have the aspects that we associate with even the most simple forms of intelligence, namely self-awareness and free will. Think about “inferior” intelligences that we commonly interact with, cats and dogs. Would anyone deny that both animals have self-awareness? When we get a computer program that can demonstrate even the rudimentary capabilities of a dog or cat, then I will start listening to the drumbeat of paranoia regarding out of control artificial intelligences running amok.

    However, I am not holding my breath.

    1. about 30 years ago someone prominent said in a keynote address that artificial intelligence has been a promising technology for the last 30 years, and 30 years from now it will still be a promising technology. people kind of chuckled, but 30 years have elapsed and things haven’t changed. having spent the last 40 years of my life attempting to simulate human thought i can tell you that we have nothing to worry about. for beginners, the human mind is not a von neumann machine. and trying to solve the problem with computer scientists will get us nowhere as has been proven over and over again.

      1. While human level AI was not achieved over the last 30 years, its incredible that you have been unaware of the advances.

        People are alive today who saw the first room sized computers to simple arithmetic and now you can talk to your phone. Even if computers are only half-way to being as intelligent as us, it is approaching rapidly and in plain sight, and a year later they will be smarter.

        1. reducing the size of computers is not an advance in ai. and the reason why a lot of people worry about ai, like musk, is that they think power of computers and ai are linked. they aren’t. the chess computer programs that everyone talks about are nothing more than an attempt to make chess into a big tic tac toe game using brute force. there is no intelligence there. doing ai is not the same thing as inventing steam engines to drive piles, but most people assume they are equivalent, we just need bigger, faster computers.

          1. Not “just” need bigger, faster. But it’s part of the foundation.

            Bigger brains are necessary for greater intelligence. Houseflies cannot be intelligent.

    2. I think the problem Mr. Musk is alluding to is that we will rely on “artificial intelligence” long before such a construct comes close to achieving real artificial intelligence.

      Call it “computer intelligence” for now, since we already turn over control to computers in areas where human intelligence isn’t well suited, like factory floor operations (boring, repetitive, error-prone) or military defense systems (speed of response, accuracy).

      And I think we are all aware enough to know that oversight will not keep this genie in the bottle. Someone somewhere with the future equivalent of a Raspberry Pi will release a version of computer intelligence that is good enough but not subject to oversight. If it is sufficiently advanced, it won’t need much human assistance; but I think it’s more likely that a sufficiently advanced computer intelligence will be scooped up by ?Chinese? manufacturers or ?South Korean? imitators to differentiate their product (“The Next Big Thing Is Here”) and we humans will be in a control battle like never before.

Leave a Reply to C++ Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.