The Singularity is Not Near

Blah blah blah singularity blah blah machine AI blah blah the world will undergo a paradigm shift, it’s coming, all bow down before the mighty new technologies that will change humanity forever. The problem I have with talk of the technological singularity is not that it doesn’t make sense, and not that I don’t believe that technological advancement is indeed rapid, accelerating, and world-changing, but that we have somehow invented a symbol of vast but actually rather vague significance. I don’t think the “singularity” is a useful idea. I think it’s a buzzword to some, and a religion to others.

For what makes Futurology (capitalization mine) really, actually different than a belief that something momentous will happen in 2012, when the Mayan calendar wraps around? Not a lot, as far as I can tell. And now it turns out that two religious scholars have concluded exactly the same thing, in a 2008 paper in the Journal of Contemporary Religion:

Futurology-as-religion has charismatic leaders, authoritative texts, mystique, and a fairly complete vision of salvation. Futurology is, in effect, a new religious movement (NRM).

Let’s break this down a little further. How will we recognize when the “singularity” occurs? Some accounts speak of a period of “unprecedented technological progress” or an exponential growth in computing power, but we’ve been seeing that for 50 years. Or it is described as a point beyond which change is so rapid that prediction is impossible, but prediction more than a few years into the future is impossible anyway, if for no other reason than the butterfly effect. It’s not that I dispute the core argument that technology will continue to alter humanity in nearly unrecognizable ways. In fact, I find many of the future technologies discussed by the Singularists to be quite plausible, including nanotechnology, better AI, and greatly extended human life — there’s lots of serious research going on in all these fields. And I also suspect that we will continue to live through a period of accelerating technological capability, because it is the nature of technology to build upon itself. But why must there necessarily exist some special point? And why must all of these technological transformation necessarily be, well, good?

Even quantitative exponential growth in computing power (or other measurable human capacities) doesn’t imply a singularity. Exponential growth accelerates endlessly, but not infinitely fast; it has no special points or infinite asymptotes.

The only claim that seems at all concrete, the only thing that might give a definite date to the singularity, is the moment when a machine becomes smarter than a human. Such a machine, it is claimed, could improve on itself in a recursive and accelerating fashion, rapidly exploding up to incomprehensible intelligence levels and coming to rule the universe. Surely, this would change history in Godlike ways.

Except that nobody knows what machine intelligence is. Or how we’d recognize one if we met it. The word “intelligence” suggests that one day the computer would wake up and talk to us (presumably, through IM) but this is mere metaphor. (Kurzweil et al. also speak of replacing neurons with hardware or software to produce a synthetic human brain, but that would be a re-implemented human intelligence.) The phrase “machine consciousness” is even less useful, because we can’t even define the word “consciousness” for humans.

Nobody knows what the words use to describe the singularity actually mean.

If no one can specify criteria for noticing when this singularity has actually occurred, I argue that it doesn’t exist even in a theoretical, conceptual sense. In a practical sense it’s therefore no better than Nostradamus, or 2012, or tea leaves. What’s left in the concept is merely belief: belief that somehow, somewhen, something big and important is going to happen. The End of The World (as we know it.) The Ascent to Paradise. Living forever in the consciousness of the machine. Apocalypse. Salvation.

All hail the prophet Kurzweil.

10 thoughts on “The Singularity is Not Near”

  1. Mentifex: your little word association toy is not machine intelligence. It cannot learn, for one thing. And you couldn’t use it to execute even well-understood AI tasks such as chess playing or language translation.

  2. The http://code.google.com/p/mindforth/wiki/NewConcept module of the AI Mind (in Forth or JavaScript) achieves machine learning by creating a new concept from any unrecognized English word. The AI Mind can not play chess because the AI is still very primitive and is still growing its ability to engage in self-referential thought. Language translation will be possible in mentifex-class AI minds by virtue of the fact that each AI Mind is capable of being programmed with both an English lexical array and, say, a German lexical array. Then thinking in the deep Psi concept array will be expressed in either English or German, or both during translation and interpreting. What looks like a “word association toy” is indeed machine intelligence and has already spawned http://AIMind-i.com as a branch in the evolution of machine minds. So there 🙂

  3. Everything in this post is true, thanks for posting it

    To continue with the religious comparison, which is very accurate, the singularity is much like the second coming or the rapture – a blissful time which shall be visited upon all who believe, for which you don’t have to do anything. Of course no one can know what it is because it is beyond mere human intellect.

  4. the singularity is different from traditional religion although i do not deny that it is the newest incarnation of a human desire for immortality and some religious concepts, but the true difference is that no transhumanist or person believing in this singularity can tell you with out doubt of what specifically will happen because even in this concept a giant meteor destroying the earth is possible and there will be no divine intervention to save humanity if its destroyed before the singularity. also no one intelligent any way thinks it will be a perfect Eden with out issues because technology is a double edged sword and there will always be issues. have you seen they have robotic arms and legs controlled by the mind and nerves they have implants so the blind can see mind reading helmets you can buy i would say we are in the singularity admittedly in the early stages the iPad is just first generation external mental enhancement. i think the question isn’t when machines beat human intelligence but rather when do we decide invasive implants and physical advancement is acceptable because im sure in some respects its not yet. also i personally believe the first true AI will use neurons on micro chips and be sort of a have biological half mechanical being that actually in some respects has a soul because each chip will be different due to the neurons. they have done this experiment on rat brains and neurons each set of neurons behaves differently and can think and learn. when we perfect this i think we will have crossed a boundary you say is impossible to quantify.

  5. The post is well organized. I visit the author has a real knack for this topic.
    I like that theme, and I’m in constant search of new
    pieces and the latest news. I truly loved that one, as it’s full of interesting facts and it’s a sort
    of article. I spent just a couple of minutes reading, and because of well-structured
    text, so I know it totally. Thanks!

Leave a Reply

Your email address will not be published.