Hyperintelligent Machines: Myth Or Menace?

Here’s a provocative item from Wired: a skeptic’s take on the idea of superhuman AI, by one Kevin Kelly. I haven’t time to comment on it now, other than to say that once you get past the “there are all kinds of intelligence” boilerplate, it raises some interesting points.

Related content from Sphere

3 Comments

  1. Bluefin Tuna says

    Being supplanted by “unstoppable” AI is difficult to imagine. I have been beaten many times at chess by a computer, but I am quite sure I could have tamed it by pulling out the plug, or at least by hitting it vigorously several times with a hammer. True, one might design a robot capable of fighting back, but only if one wanted to use it as a weapon against one’s enemies- and in that case, programming it to destroy oneself along with them would seem quite counterproductive.

    His point about costs and materials is a big one. I recently heard about a FOAF who was involved in testing a robotic pack mule for the US Army, for use in rough mountain terrain impassable to conventional vehicles. After many experiments, they realized that the cost of building and fueling the robot was inevitably several orders of magnitude higher than that of simply buying and feeding a flesh-and-blood donkey. Hundreds of millions of years of evolutionary fine-tuning have made animal cells far more fuel-efficient and durable than any man-made vehicle on Earth.

    Most fundamentally, genuine self-consciousness is a basic human advantage that is almost certainly impossible to replicate with AI. You can simulate the outward signs and appearance of self-consciousness, of course, but no matter how sophisticated, you still have nothing more than a clever version of Searle’s “Chinese Room”. A robot may fight against an attacker because it has been programmed to do so, but not because it fears death, values its own life, or is in any way an egotist.

    Posted June 28, 2017 at 12:42 pm | Permalink
  2. JK says

    http://dailytimewaster.blogspot.com/2017/06/it-is-claimed-that-robots-will-take-all.html

    Posted June 28, 2017 at 5:30 pm | Permalink
  3. Malcolm says

    BT,

    A robot may fight against an attacker because it has been programmed to do so, but not because it fears death, values its own life, or is in any way an egotist.

    I don’t think this is a valid distinction. Those things are just ‘programmed’ into us as well – and all of those responses can and do proceed quite independently of consciousness, even in humans.

    Posted June 29, 2017 at 1:32 pm | Permalink