Upload Skepticism

David Linden in Boing Boing:

Kurzweil predicts that by the late 2030s, we will be able to routinely scan an individual’s brain with such molecular precision and with such a complete understanding of the rules underlying neuronal function and plasticity that we will be able to “upload” our mental life into a vastly powerful and capacious future computer. … I am a neurobiologist and I have spent the past 28 years engaged in studies of the cellular and molecular basis of memory and cognition. I am an optimist and a technophile, but I believe that I speak for the vast majority of brain researchers when I express serious doubts about Kurweil’s timetable. …

Kurzweil then argues that our understanding of biology—and of neurobiology in particular—is also on an exponential trajectory, driven by enabling technologies. … At some point in the 2020s, a miracle will occur: If we keep accumulating data about the brain at an exponential rate (its connection maps, its activity patterns, etc.), then the long-standing mysteries of development, consciousness, perception, decision, and action will necessarily be revealed. … That’s where I get off the bus.

Our understanding of biological processes remains on a stubbornly linear trajectory. … There have been a number of genuine paradigm-shifting insights in genetics in recent years. … But these discoveries, and most of the other key conceptual breakthroughs in this field, have come slowly, the result of stubbornly linear small science, and not of the huge technology-driven data sets that Kurzweil describes. … This linear progress also holds true for the growth in our knowledge of brain function. … The ploddingly linear increase in our understanding of neural function means that an idea like mind-uploading to machines being usefully deployed by the 2020s or even the 2030s seems overly optimistic. (more; HT Tyler)

I’m happy to defer to Linden’s brain science expertise. But I wish he’d get clear on two key points:

  1. All this talk of linear vs. exponential progress, with linear progress unable to finish by 2040, suggests Linden has in mind a rough estimate of how far along are we now, and how fast we have been proceeding. It would be very helpful if Linden would tell us his best guess. For example, are we now 20% of the way along, and progressing 5% per decade, suggesting we need 160 more years of linear progress?
  2. Linden talks about “mysteries of development, consciousness, perception, decision, and action.” But all we need for brain “uploads” (= emulations) are good enough models of individual cell input/output/state relations (i.e., how a brain cell’s output signals and internal states change as a function of its input signals). We don’t need to understand how that huge mess of connected cells actually produces high level brain functions. If we just focus on this more limited goal, then how far along are we, and how fast are we moving?

Yes Kurzweil seems too optimistic, but rather than criticizing Kurzweil it seem far more useful for Linden to offer his own best expert estimates. Brain emulations would have such enormous social implications that even if they will take a century or so to arrive, it is still very important to let people know, so we can start to prepare. I fear my economist colleagues will continue to ignore this possibility until top brain scientists like Linden tell them it really is coming.

GD Star Rating
loading...
Tagged as: ,
Trackback URL:
  • candy

    Typo: “Kurweil”

    Kurzweil, I think, imagines that our neuro techniques will keep up with the hardware, and predicts a $1000 human-equivalent computer in 2029 following his trend lines. I’m not so confident about the neurosci but I do think that kind of computing ability will allow big changes even if the brain is still a bit mysterious.

  • Jamie_NYC

    Robin, you will never get the answers you are looking for from Linden and people like him. He is simply expressing his weltanschauung, not making any predictions. The gist of his worldview is ‘never’, or, more concretely, the things Kurzweil is talking about will happen in a future so distant that it is indistinguishable from ‘never’ from a personal perspective.

    It seems to me most people, even technologically sophisticated, are like that. Why – perhaps that’s a good topic for you to discuss.

  • Dustin

    These were the exact thoughts I had when I read the article at Boing Boing earlier.

    I also think Jamie_NYC is spot on.

  • http://elbizri.com Jon Bizri

    I had the feeling, when reading that article, that the concept was ‘too silly’ for him to think about.

    http://www.overcomingbias.com/2008/04/arbitrary-silli.html

  • Buck Farmer

    If we had all “individual cell input/output/state relations” sufficiently modeled and could then scan an individual’s full nervous system…

    How would we know what inputs to feed the modeled system?

    It seems we could very easily end up with either “brain-dead,” “insane,” or “perpetually tortured” EMs unless we have a pretty comprehensive and detailed picture of how the non-neurological world hooks up with the neurological world.

    Take something fundamental like your nervous system’s ability to know that your organs aren’t dying, being punched, poisoned…unless we know how the “everything okay” feedback loops work, it seems more likely we’ll end up with insane EMs than not.

    On top of that…it would probably advance progress towards non-insane EMs to have millions upon millions of insane ones to test.

    By most reasonable utilitarian or deontological ethical systems, this sounds like a tremendous amount of evil.

    • http://lukeparrish.rationalsites.com/ Luke Parrish

      It seems we could very easily end up with either “brain-dead,” “insane,” or “perpetually tortured” EMs unless we have a pretty comprehensive and detailed picture of how the non-neurological world hooks up with the neurological world.

      I find it hard to believe that an EM’s pain or insanity would be something we could not determine by fairly casual inspection and thus halt the simulation within a few of the EM’s subjective seconds. Only deliberate cruelty or complete apathy would result in the kind of scenario you are predicting. Pain is a very strong signal that commandeers the attention of the conscious mind and produces recognizable reactions.

      Note also that EMs can be (and in the beginning will be, due to hardware limitations) run much slower than biological entities, so one could study a second of their time for days or years on end. This is part of what makes them so attractive for studying — slow motion and reversibility, combined with the ability to access all of the data without scanning, makes it much more likely that we can reverse-engineer the essential algorithms and thereby avoid the ones that cause instability or distress.

    • http://www.gwern.net/Links gwern

      > How would we know what inputs to feed the modeled system?

      Why do you care? The brain has demonstrated tremendous flexibility. Eyes using your tongue, spatial perception using electricity on your head, geographic orienting and navigation using a buzzing device, echolocation by clicking, goggles that invert your vision and eyes that adapt to that in hardly any time… Just this writing and your reading demonstrates how flexible the symbol-processing of the brain can be with practice!

  • Drewfus

    That’s where I get off the bus.

    No determinism for this man, thanks. Linden’s life has quality, and people like Kurweil aren’t going to take that away from him. Behavioristic knowledge and technology growth might be as dehumanizing for Linden as being asked to participate in a time and motion study.

  • http://opines.mythusmage.org Alan Kellogg

    Kurzweil’s mistake lies in his understanding of personality. He thinks it is a static thing, where it is actually a dynamic process.. You can get a snapshot of a personality, but all you upload is that snapshot. You don’t get the person in question, all you get is a scan of that person, and thus a copy.

    Then you have what happens to that scan in the limited venue of a computer. For a personality is not only the product for a memory, but of the interaction of a human mind with an external world. Will we ever be able to replicate reality to any degree with currently foreseeable computer technology. Or will we have to depend on a miracle to produce the miracle needed.

    We’ve got a lot more to learn about human minds and reality before we can confidently talk about uploading personalities to some predicted future.

    • http://lukeparrish.rationalsites.com/ Luke Parrish

      You can get a snapshot of a personality, but all you upload is that snapshot. You don’t get the person in question, all you get is a scan of that person, and thus a copy.

      This so-called “snapshot” would change afterwards based on the results of the simulation. The primary difference is you could create a backup of the digitized personality at any given point in time, without having to go through all that messy business of scanning them.

      Then you have what happens to that scan in the limited venue of a computer. For a personality is not only the product for a memory, but of the interaction of a human mind with an external world.

      Most of my imagination has no direct interface to the physical world. I have to use crude implements like my hands, voice, eyes, etc. to manipulate and perceive reality. Replicating (and exceeding) that degree of interactivity would fall quite short of miracle territory.

      Would it be me? Debatable. Would it be a person? Not so much.

  • Ari T

    Clearly you need a prediction market.

    Honestly though, Kurzweil is way too optimistic. I’d say nuclear war is more likely than this, although I wouldn’t say any serious predictions unless I were a neuroscientist. People from other professions tend to overestimate practicality of any “cool” technology. So many people have predicted the revolution of speech recognition but the practical problems keep piling up. Kurzweil has been too optimistic on most of his predictions. And he doesn’t take into account the inertia created by government inefficiencies, direct and undirect.

    If a neuroscientist is stating serious doubts about these predictions, I’d say the reality is even more pessimistic, given that usually technology’s researches are the ones that are too optimistic. Usually there tends to be a myriad of practical problems

    And I think the reason people are not so interested in this is the same reason people are not really interested in investing or betting on the issue. There’re too many practical issues.

    p.s. I think from ethical point of view your discussion with Karl Smith was interesting though.

  • donK

    It’s generous of you to defer to an brain science expert when you so clearly understand what “we need for brain ‘uploads.’” On the other hand the future may be shiny and your armchair science misses many details that require more than a miracle.

    • http://lukeparrish.rationalsites.com/ Luke Parrish

      your armchair science misses many details that require more than a miracle.

      Such as?

  • http://www.google.com/reader/shared/saliency Saliency
    • Dan Weber

      They caught him stealing paperclips. Rather, his em was maximizing them through theft.

  • hf

    But all we need for brain “uploads” (= emulations) are good enough models of individual cell input/output/state relations

    So you say. Others* argue we’ll need to understand what the parts actually do in more detail. This would obviously make non-upload AGI more likely.

    *Namely, everyone else I’ve seen discussing the issue, though that does seem like a small sample size.

  • Dave

    Wait, let’s upload colons first.
    What is always so puzzling about these discussions is they are based on unexamined assumptions and mental shortcuts. The one thing these EMs will have is intelligence and awareness.But we have no idea of what awareness consists. No one knows how to enumerate the phenomenon of “intelligence” Is it being good at math or an ability to understand complex metaphors? Computers are great at math but absolutely stink at understanding metaphors.

    So we are going to build a computer that is so powerful that it can crush this problem with its sheer computing power. This is the province of computer nerds and sci fi fans. It partakes of the fallacy of infinite extrapolation plus inept comparisons between digital electronics and biologic systems.

    Biologically the brain has more in common with the colon than a computer.The colon has a complex nervous system that reacts to biologic stimuli and social stimuli as well as hormonal and chemical influences in an all but unconscious but very organized manner. Every normal person has one whether a genius or a dolt. I Suggest we take things step by step. before indulging wild claims speculation about EMs.

    I propose that we begin a crash program so that by the late 2030s, we will be able to routinely scan an individual’s colon with such molecular precision and with such a complete understanding of the rules underlying neuronal function and plasticity that we will be able to “upload” our colons life into a vastly powerful and capacious future computer. By starting with a simple system we can then we can proceed through the other organ systems until we can upload the brains of creative geniuses for the benefit of all humanity.

    • Cyan

      Frankly, this is a great idea just for what it would contribute to medical research, never mind upload tech. Something like your crash program, perhaps on a smaller scale, already exists under the name “systems biology“.

      • Dave

        Interesting. Of course I was partly kidding. However I have spent my life in the biologic/medical area. There is not a complete account of the way the basic organs or even cells interact.Since they act silently and with little influence from the conscious mind they don’t get much respect. It is not clear to me that they are any less sophisticated than the human brain itself which in many ways is the most poorly perfected organ. The second most imperfect organ is the skin.

      • http://daedalus2u.blogspot.com/ daedalus2u

        It is likely that they are more complicated, not less. But humans are in the habit of paying attention to squeaky wheels, and then thinking that what they pay attention to is what is most important.

  • http://www.hopeanon.typepad.com Hopefully Anonymous

    Yes Kurzweil seems too optimistic, but rather than criticizing Kurzweil it seem far more useful for Linden to offer his own best expert estimates.

    Huzzah, Prof. Hanson. Reminds me of expert critiques of Dr. Aubrey De Grey, a few years ago.

  • komponisto

    Linden writes:

    I am an optimist and a technophile, but I believe that I speak for the vast majority of brain researchers when I express serious doubts about Kurweil’s timetable.

    This is confusing. Is he saying that he himself is optimistic, but is obliged to report the contrary consensus of his colleagues (which possibly reduces his confidence in his optimism), or that his colleagues’ views are so persuasive that even he, “an optimist and a techophile”, agrees with them?

    If he is unusually technophilic among brain reseachers, then it isn’t unexpected that in speaking for the rest of them (as opposed to himself) he would be more pessimistic; hence the word “but” is inappropriate. (He should have said: “I’m an optimist and a technophile, so it’s for the rest of my profession and not myself that I’m speaking when I express doubts about Kurzweil.”) On the other hand, if he, as a technophile, agrees with his non-technophile colleagues, then the rhetorical emphasis should be on his own unexpected pessimism, not the expected pessimistic consensus, and he shouldn’t need to point out that in expressing pessimism he “speaks for” other (non-technophile) brain researchers (whose consensus can already be taken for granted to be on the pessimistic side).

    It would be an appropriate rhetorical formula for a member of the non-technophile consensus, against a technophile: “Linden may agree with Kurzweil — after all he’s an optimist and a technophile — but I believe I speak for the majority of brain researchers when I say that Kurzweil is wrong.” Not for a technophile himself unexpectedly agreeing with the consensus. Such a person should rather say “I’m an optimist and a technophile, but even I agree with the standard pessimism of brain researchers.”

  • Ian

    Personally, I’d say we need to focus more on making the future than predicting it…

    ~Ian