Entries:
Comments:
Posts:

Loading User Information from Channel 9

Something went wrong getting user information from Channel 9

Latest Achievement:

Loading User Information from MSDN

Something went wrong getting user information from MSDN

Visual Studio Achievements

Latest Achievement:

Loading Visual Studio Achievements

Something went wrong getting the Visual Studio Achievements

Discussions

Bass Bass Knows the way the wind is flowing.
  • Artificial intelligence could end mankind

    @JohnAskew:

    Unfortunately I view religious belief as socially acceptable psychosis. It's strange to me that belief in the tooth fairy or Santa Claus is considered silly, but the ramblings of a character in a ancient sci-fi novel is considered sacrosanct by apparently "mature" adults.

    It's unfortunate that most arguments I've encountered against the possibility of AI tend to boil down to some fundamental belief in supernaturalism, and it shows that religion continues to corrupt thought processes the world over.

  • How would you defeat this argument against open source?

    , evildictait​or wrote

    And the concept that open-source software is free of control by a single entity is equally ludicrous. I can't just submit a patch to Linux and get it pushed out tomorrow. That patch needs to be accepted by a collection of open-source programmers who control the Linux kernel. I can't add a new tag to Mozilla's rendering engine and push it out. It's open-source, but the code is still controlled by Mozilla Corporation.

    Sure you can. git pull from Mozilla, git commit, git push to github. You don't have to have permission to do that. In fact, many people don't use repos from Mozilla or Linus directly. When you use Red Hat or Ubuntu, you are probably using THEIR Linux kernel and THEIR Firefox which derive code but aren't equivalent. That way they don't have to rely on upstream to push security patches. I do this too regularly.

  • How would you defeat this argument against open source?

    @evildictaitor:

    Why bother with that tool? Why not use the vuln you found to break into company's systems, steal the source code of the vulnerable product and patch it/release a patch yourself? Because that's illegal, right?

    The difference between proprietary and open is a legal distinction. You are opening yourself to copyright infringement liability if you start modifying someone else's software in which the license does not permit you to do so. Even for altruistic purposes. 

  • Artificial intelligence could end mankind

    , JohnAskew wrote

    @Bass:Science fiction.

    "we don't really know how" sums it up. Artificial Intelligence is a de facto oxymoron.

    It will always come up short.

    Intelligence cannot be artificial. And artifice cannot produce a sentient machine.

    Why?

    Sentient, that's what AI purports itself to be. Play chess all you want, that's not intelligence, it's strategy from data. Programmers determine the closed system to play the closed game of chess, how will that algorithm think or procreate? It can only vomit. Asinine concept, AI, farcical, petulant, silly, the stuff of science fiction. 

    Machine learning techniques essentially learn an algorithm from data. The programmer in this case, doesn't tell the machine to do anything. He gives it a goal and a way to learn.

    Sentient beings have an a-temporal element about them which can never be approached with computer algorithms or temporal physics. The entire idea of AI is laughable and those who exalt it as something it can never be have some other agenda in mind (i.e. Musk).

    Are you making some kind of religious argument?

  • How would you defeat this argument against open source?

    @KirbyFC:

    The advantage of open source is anyone can fix it, not that you personally will fix it. With proprietary software, only the proprietor has the legal allowance to produce derivative works of the software. This is not so with open source. Essentially, the right to modify and redistribute is one of the only concrete distinctions between open source and proprietary - the distinctions relate not to technology but the permissions in the license.

    That does not guarantee someone will fix it - especially if it is not a popular piece of software. The only promise that open source provides is that someone other then the proprietor can legally fix it.

  • Artificial intelligence could end mankind

    I want to add my personal opinion is the breakthrough leading to AI in optimization. An efficient way to minimize some function with respect to its output. Sounds easy right? It's really * hard, but it's the key thing driving everything.

    Some of the stuff happening in 2014 strongly hinted towards the idea that we don't really understand how to do optimization (and somehow we still do really powerful things with our shitty understanding). We have powerful neural networks, but without the right optimization algorithm, they are not as powerful as they can be.

    Btw, by bad AI, I don't mean The Terminator. That's silly.

    I mean an unstoppable superintelligence that if designed incorrectly will consume all useful matter and energy in the Universe in a rate that converges towards the physical maximal possibility.

    You've seen pretty bad bugs right, you know like a bug in a critical system that cascades failures in other related systems? Badly designed AI could segfault the whole f**king Universe. Nobody will be around to file the bug report.

    It could even start off with the intelligence of a gnat and end up like this quickly. Hollywood's imagination is limited, and they require silly things like weaknesses humans can exploit. You need that happy ending to sell tickets!

  • Artificial intelligence could end mankind

    , JohnAskew wrote

    @Bass:sources, please.

    Sure. Here is a selection of some papers I felt were milestones towards AI in 2014. A background in statistics and machine learning is useful:

    http://arxiv.org/abs/1410.5401

    http://arxiv.org/abs/1409.4842 

    http://arxiv.org/abs/1312.6184

    http://arxiv.org/abs/1412.2306

    Other things worth looking up: "Hessian-free optimization" and "Natural gradients". Also some of Bengio's work on alternatives to backpropagation. It's hard to say if this will led to true AI, but personally I don't think we are that far away, maybe one or two significant breakthroughs away. On a more philosophical level, DNA is less than a gig of information and only a tiny fraction of that is dedicated to intelligence. Thus, you can assume that the key to intelligence is subtle but not terribly complicated or heuristic.

  • Artificial intelligence could end mankind

    , JohnAskew wrote

    The death knell of AI is when they pat each other on the back and say how powerful AI is...

    It's like a toy chihuahua barking fiercely. AI ain't all that. It's sci-fi.

    These people are respected scientists not foo sellers. The reason you are seeing high profile scientists coming up with things like this is because we believe we are a lot closer to general AI then the first sentence of that article leads you to believe. There has been an unusually large number of breakthroughs just in the last year, last few months even. People with decades of experience in the field who would normally have measured responses to the possibility of AI are getting a little worried that Kurzweil might have been off by a few decades. That's why it's important to prepare now.

  • Interesting Items from CES

    If I was forced to use RaspPi for graphical kiosk like applications, I'd use probably [attempt] to use Clutter (straight C code) and Wayland or the framebuffer directly. Or even drop Clutter and just go direct OpenGL. I definitely would not try to bring the whole X11 windowing system and a web browser along for the ride.

    But IMO, unless you are shipping a million units or something I think it just makes sense to use a high level language/framework you are productive in and buy a better computer. :) Cost of developers >>> cost of hardware, most of the time!

  • Interesting Items from CES

    , Bas wrote

    *snip*

    I'm so glad to finally hear someone who agrees. I need a sort of cheap kiosk solution and everyone keeps going "hey, just use a pi, it's perfect!" but it is just too ridiculously slow. I'm not sure why everyone keeps suggesting it as the one size fits all solution for everything. In the end we just got a bunch of NUCs running some sort of kiosky Ubuntu version.

    I could have lived with the 512MB RAM. But the CPU is really weak, and that's the thing I've noticed that web apps crave. People see 700MHz as not that bad thinking back to the PentinumIII but they are thinking in terms of Intel's crazy superscalar processor designs with all kinds of fancy branch prediction and cache prediction logic they use and used to really get the most out of that 700MHz on general purpose workloads. ARM has only barely started to compete with Intel on things like that, so I'm not even so sure about the improved Pi competitors. The GPU isn't that bad though, but you really have to be running Wayland to get the most out of it.