Slashdot Mirror

← Back to Stories (view on slashdot.org)

Elon Musk, Others Fund $1B Non-Profit To Advance AI Research, Ethics (openai.com)

Posted by timothy on from the first-prove-that-an-AI-isn't-running-it dept.

As reported by The Wall Street Journal, The Verge, and many others (including this widely run Reuters story), a heavily backed non-profit group called OpenAI on Friday introduced itself and its utopian-sounding goals of open sourcing a great deal of AI research and, as The Verge puts it, "to stop AI from ruining the world." Elon Musk and Peter Thiel are two of the backers, along with Y-combinator president Sam Altman and other Silicon Valley luminaries, so the group starts out with a war chest big enough to support a wide range of research -- a billion dollars. According to the Wall Street Journal, The idea for OpenAI crystallized last summer as a result of ongoing discussions between Mr. Musk and Mr. Altman over the future of AI. Mr. Musk has warned in the past that the future of the Earth is at risk if AI develops in the wrong ways. Tesla, one of the companies he leads, is adding autonomous capabilities to its cars that require AI technology such as image recognition. ... OpenAI intends to collaborate with the academic and for-profit worlds, but it also wants to give researchers the freedom to pursue lines of enquiry without pressure to achieve an immediate pay-off. “Our focus is not only doing the right thing for today, but also doing the right thing for 50 years from now,” Mr. Brockman said.

27 of 56 comments (clear)

  1. Apparently... by burtosis · · Score: 3, Insightful

    Elon must love terrifying himself with all his talk of AI lately.

    1. Re:Apparently... by Anonymous Coward · · Score: 2, Informative

      Elon is a toolshed that goes out of his way to speak on topics to which he has no education. AI is so far away from the Hollywood doomsday picture he's painting that it's absurd to even worry about it.

    2. Re: Apparently... by Anonymous Coward · · Score: 1

      Whoever you are, thank you. As a researcher in AI undoing damage from people like Elon Musk and Stephen Hawking is a monumental task. These people are smart in their own right, but know jack shit about AI and should shut the fuck up about it.

    3. Re:Apparently... by LifesABeach · · Score: 1

      Elon has publicly stated that AI is bad, but it appears that AI is not bad if Elon does it?

    4. Re: Apparently... by khallow · · Score: 1

      . As a researcher in AI undoing damage from people like Elon Musk and Stephen Hawking is a monumental task.

      [...]

      but know jack shit about AI and should shut the fuck up about it.

      It would help if you knew more than jack shit either. None of us know what AI could do. All we have are some incredibly weak thermodynamic constraints.

  2. Peter Thiel is a lunatic by Giant+Electronic+Bra · · Score: 1

    I don't know how that guy managed to make his money, luck basically. He's got the good judgement of a pigeon.

    --
    "Malo periculosam, libertatem quam quietam servitutem." -- Jefferson
  3. Re:Pattern matching is NOT A.I. by jeffb+(2.718) · · Score: 1

    Then again, good luck proving that you are.

  4. Hold On Tight! by JimSadler · · Score: 4, Informative

    Computers, 3D printing and AI are all super, major developments. Any major development in human history has carried a certain amount of chaos, pain and suffering along with it. AI is going to cause a total shift in the nature of human affairs and existence. But in theory, we have an advantage. We can see it coming and we can buffer the negative consequences if we choose to do so. Obviously human employment will become quite rare in the near future. People without jobs must be well supported by governments. They must have all that is needed to get by plus more so that they can consume products produced by industry. Taxes can only come from people who can afford to pay them and that translates into the wealthy and businesses paying the tax load for the rest of us. I know the knee-jerk reaction that some people will have when i mention this. And just maybe that is the area that we need to spend money on. We need to make people aware that a huge change is upon them and that there is no choice but to eliminate many long-held beliefs in order to have a smoother transition into the new era. Imagine the chaos that went on when we went from the horse to the car. That was nothing compared to what Ai has in store for us.

    1. Re:Hold On Tight! by DNS-and-BIND · · Score: 1

      What, again? We have to drop everything and adopt socialism NOW? This argument comes up every single time. The arguments are all different and they are worded in different ways but the ideas are the same and the last one is as wrong as the first."

      You don't even know what socialism means. It means work. People who don't work...socialists will coldheartedly let them starve. Yup! For realsies! Check into your own history.

      --
      Shutting down free speech with violence isn't fighting fascism. It IS fascism!
    2. Re:Hold On Tight! by dinfinity · · Score: 1

      Respectively: no, no, and the (terrible) execution of it does not define socialism.

      More importantly: You entered the term into the discussion. You chose to use the most dire depiction of it as the definition. Nothing in Jim Sadler's post implied that we should let people starve. If that is what socialism is to you, then obviously Jim Sadler was not talking about 'socialism'.

      In simpler terms: you created and attacked a straw man. A classic and egregious fallacy in (your) reasoning.

    3. Re:Hold On Tight! by rtb61 · · Score: 1

      Do not confuse any socio-political system with the actions of individuals. No matter the socio-political system psychopaths, will corrupt it every single time in their favour and collapse the system over time. Communism, capitalism, socialism, democracy, anarchy, all collapse under the corruptive influence of psychopaths, simple fact of life.

      For artificial intelligence the focus should be on application. So most current application, translation services. So automatic written translation in proper context from one language to another. Second step spoken language recognition and translation, from one language to another and of course as part of that from human language to computer language, something that a computer can reliably interpret to carry out actions. This all require a learning computer ie learning written and spoken mannerism and language usage. So AI with very limited function, basically the core of AI development, not thinking for us, simply contextual understanding of instructions.

      Obvious benefit to humanity from this kind of AI, every one being able to talk to everyone else, exchange ideas, minimise divisions and focus on our real problems and the psychopaths at the top causing them. Breaking down communications barriers as is already being demonstrated is changing our world. This being exemplified by the steady shift to social democracy from other forms of government and this in conjunction with the recognition of the extreme harm psychopaths cause to human society (get rid of them and pretty much any style of government works effectively and with them pretty much any style of government devolves and collapses).

      --
      Chaos - everything, everywhere, everywhen
  5. A bit of a reverse Pascal's Wager? by jeffb+(2.718) · · Score: 1

    If they take these actions, and malevolent strong AI does not come to pass, perhaps they've saved us all.

    If they take these actions, and malevolent strong AI does come to pass, they're no more doomed than they would've been anyhow.

    Of course, Pascal's wager falls apart if you pick the wrong god to worship (or the wrong way to worship it), and Musk and Thiel's wager falls apart if the world-consuming evil AI has a soft spot for goodlife. And if strong AI doesn't happen in any form, maybe they've wasted their money.

    1. Re:A bit of a reverse Pascal's Wager? by PolygamousRanchKid+ · · Score: 1

      From TFS:

      "to stop AI from ruining the world."

      There is no need for AI to ruin the world . . . humans are perfectly capable of doing that themselves, without the help of AI.

      A loftier goal for these rich folks, would be to research ways to stop humans from ruining the world.

      --
      Schroedinger's Brexit: The UK is both in and out of the EU at the same time!
  6. Probably not the best use of $1B by mykepredko · · Score: 2

    I guess Musk got an AI demo that scared the bejeezus out of him because I always think about an article I read about the genesis of SpaceX where he sat down for a couple of days with a spreadsheet working at figuring out if there was a business opportunity with a reusable launcher.

    That kind of data centric, analytical approach doesn't seem to be taken here - it seems to be more knee-jerk worry than anything else.

    Better uses for the $1B? Let's see:
    - Practical electric propulsion for airliners.
    - Large scale power storage and distribution systems for EV, wind and tidal (all renewable) energy sources.
    - Lunar mining, refining and smelting to help create a sustainable industry off earth.
    - An electric delivery van.
    - Accelerating the Dragon II development.

    All would provide returns which would help mankind, would disrupt entrenched industries and provide a (somewhat risky) ROI for other investors to join in.

    --
    Mimetics Inc. Twitter
    1. Re:Probably not the best use of $1B by SuricouRaven · · Score: 1

      "Lunar mining, refining and smelting to help create a sustainable industry off earth."

      The most expensive construction ever built off-earth to create a product for which no customers exist. You'll need more than a billion to launch an entire space industry.

  7. AI... by matbury · · Score: 2

    AI is no match for human stupidity.

    1. Re:AI... by leftover · · Score: 1

      and this is the most encouraging comment around!

      --
      Bent, folded, spindled, and mutilated.
    2. Re:AI... by Kjella · · Score: 1

      Perhaps we should work on artificial stupidity instead :)

      --
      Live today, because you never know what tomorrow brings
  8. Re:Pattern matching is NOT A.I. by KGIII · · Score: 1

    Logito ergo sum.

    Of course, I could just be a figment of your imagination.

    --
    "So long and thanks for all the fish."
  9. Their AI's won't be as good as the AIs of the rich by PeterM+from+Berkeley · · Score: 1

    Your proposal to give people without jobs/wealth AIs to trade for them and make them money won't work.

    The rich will have better AIs, and the net effect will be to transfer wealth from the poor to the rich via stock trades.

    Seriously, d'you think your AI sitting on your slow computer at home is going to beat the AI that's running in a co-located supercomputer center just a few ns away from the trade house? With special hardware that allows microsecond response times to shifts in stock prices?

    --PeterM

    --PM

  10. I blame the current Dilbert storyline by joeblog · · Score: 1

    I suspect the paypal billionaires get their AI views from Dilbert

    http://dilbert.com/strip/2015-...

    --
    If it works, it's obsolete
  11. Where is he getting the government funding? by reemul · · Score: 2

    Elon Musk is a brilliant man, but he won't hand you a hanky without finding some way to get a government grant or subsidy out of it. Every one of his businesses gets at least some money shaken out of the taxpayers somewhere in the process. So where is he getting the federal or state money for this venture? I can't believe that he has broken with his long practice of finding a way for the government to pay him to do what he was going to do anyway, that would be a bigger story than the AI thing.

    --
    You're just jealous 'cuz the voices talk to *me*
    1. Re:Where is he getting the government funding? by czert · · Score: 1

      You know, governments *want* brilliant people to take advantages of their subsidies. That's the only reason those subsidies exist, actually: to steer economic potential the way the governments want to. I fail to see what's wrong with Musk taking part in that.

  12. Re:Their AI's won't be as good as the AIs of the r by dinfinity · · Score: 1

    Dude, I mined an entire BitCoin this year with my home computing cluster.
    That's right, I'm a pretty big player in the BC-scene, man.

  13. Gov. Military, Police, Terrorist are the problem by jraff2 · · Score: 1

    Committing to controllable AI is fine, but the Government, NSA, CIA, FBI and the other TLAs will ignore the research and do what THEY want and to hell with reasonable AI. Same for the Military, Police and Terrorists! You really think they will give a hoot about controllable AI. I for one would wold be on the lookout for SkyNet.

  14. Re:Cylon Musk doesn't want any AI competitors.. by Noughmad · · Score: 1

    Has anyone seen him hit another man? Has anyone seen him eat? No? Then he's a robot.

    --
    PlusFive Slashdot reader for Android. Can post comments.
  15. I'll give $300 million by NewYork · · Score: 1

    I'll give $300 million for carving out a separate country for Untouchable/Dalit in India http://wh.gov/ivsch

    --
    Casteism