Kurzweil Defends Predictions for 2009, Says He is 102 for 108.

74 18 Loading

How accurately must you predict the future before you can call yourself a futurist? That’s the central question in a bit of online controversy regarding Ray Kurzweil. The well known inventor and author has made many predictions over his career, and if you’ve seen any of his lectures you know about the major ones. Artificial intelligence will match and then exceed the human brain. Humans and machines will merge. The universe will be full of intelligence. But it’s a series of more precise predictions that he wrote for his 1998 book, The Age of Spiritual Machines (TASM) that are drawing flak. In a recent post on his blog, Accelerating Future, Michael Anissimov has challenged TASM’s predictions for 2009. Anissimov is no futurism lightweight, he’s an integral part of the Lifeboat Foundation, the Singularity Institute and the Immortality Institute. He and Kurzweil both work with the Singularity Institute for Artificial Intelligence and the Singularity Summit. In his post, Anissimov points out that seven of Kurzweil’s forecasts have failed to come true by the end of 2009. Kurzweil responded to the critique on the blog. The biggest point: TASM makes 108 predictions for 2009, and Kurzweil says that 102 were correct or “essentially correct”. This gives us all another chance to debate a favorite issue among technophiles: Does Kurzweil know what the hell he is talking about?

kurzweil vs annissimov

Kurzweil (left) has defended the predictions that were critiqued by Anissimov (right) in his blog. Intense futurist debating action!

In this corner, the challenger…

First, let’s just jump straight into the predictions that Anissimov critiques. I’m going to paraphrase horrifically here so I encourage you to browse through TASM on your own to get the original text.

1) People will be wearing computers in the form of jewelry and clothing.
2) The majority of text will be generated using speech to text software.
3) Computer displays in glasses will be able to project images directly onto the eyes.
4) Three dimensional computer chips will be commonly used.
5) Translating telephone technology will be commonly used.
6) Warfare will be dominated by airborne drones, some as small as birds.
7) Intelligent roads (really highways) will allow humans to sit back and let the computers do the driving.

As you can tell by my gratuitous use of hyperlinks, many of these technologies are in development if not in commercial use. None, however, are so widespread or dominant in their field that we can point to them and say “oh, well that’s obviously come to pass.” Like any good critic, Anissimov uses this lack of overwhelming evidence as a prod for Kurzweil and other futurists. He calls for a departure from “story telling” futurism towards one that focuses on probability distributions for events. Anissimov asks for all of us science minded folk to hold our futurists to strict account. If we don’t police these wild theorists, who will?

And in this corner, the champ…

Well, Kurzweil will. In his retort he carefully points out the ways in which he has already pursued the efforts that Anissimov is advocating. His predictions in TASM were for decades, not particular years (2009 to 2019, 2019 to 2029,etc). Most of his forecasts should be seen as having a plus or minus probably distribution around them. Kurzweil is also writing his own report to review and analyze his work in TASM. According to his response in Accelerating Futures, that report is forthcoming. In short, the futurist is not giving us magical exact dates, and is critical of his own work.

Of course, he also illustrates the ways in which Anissimov fails as a critic. Grabbing 7 out of 108 predictions and using them to declare a failure is misleading. Even if all seven are wrong (more on that below) what about the other 101? Personally, I’m impressed with anyone who has greater than 90% accuracy in future forecasting. Kurzweil also questions whether being off by a few years is really failing to predict the future. If all seven come to pass in 2012, was Kurzweil an idiot? By asking for probability distributions to predictions, but then demanding precision in those predictions to qualify for success, Anissimov is venturing into (but not reaching) hypocrisy.

Much as he did when challenged by Daniel Lyons in Newsweek, Kurzweil also takes the opportunity to actively defend his predictions one by one. Again, I’m paraphrasing grotesquely, so feel free to jump to Accelerating Future and read the original responses:

1) iPod shuffles, health monitors, hearing aids…these are all computers that can be worn on the body and fit the prediction.
2) Speech to text is gaining ground. It’s available in hand held devices, and as (semi) popular Apps on smartphones.
3) The technology to project images on the eye from a device embedded in glasses does exist. It isn’t common yet.
4) By 3D we mean any layering of silicon, not necessarily cubes. Such 3D computer chips are being shipped for sale now and becoming more common every day.
5) Like speech to text, this technology is gaining ground and has related Apps for smartphones.
6) Drones are a major part of the war in Afghanistan. Some of their explosive ordinance is the size of birds and contain their own navigating intelligence.
7) Kurzweil doesn’t address this one directly. I’d agree with Anissimov that we can consider it largely a “fail”, but just for now. The next few years could prove Kurzweil correct.

Even looking just at these cherry picked predictions from TASM, I’m going to give this debate to Kurzweil. If we included the other 101, I’m inclined to think that the futurist will win there as well. We’ll have to wait for Kurzweil’s essay on the subject and see how he interprets his work in TASM. According to his blog response the score (in his thoughts) is: 89 correct, 13 “essentially correct”, 3 partially correct, 2 are ten years off, 1 is just wrong (but it was tongue in cheek anyway).

Again, I find 90% or greater accuracy to be remarkable. Of course, when reviewing your own work it’s easy to see how you were right. I look forward to how others will score Kurzweil after a careful examination of all 108 predictions (starting with making sure there are really 108 of them). If most of these are accurate, what does that mean for the future?

Not much. Sure, it may mean that Kurzweil is extraordinarily gifted at seeing how trends will develop in the future. Yet those predictions in of themselves are not what causes something to happen. It takes the combined work of the global scientific community to create the steady rising tide of technology that affects our lives. At best, futurists can provide us with the insight to avoid upcoming dangers. Bioterrorism, nanotechnological pollution, anti-technology fundamentalism…these are potential conflicts that Kurzweil, or others, are actively seeking to avoid. That’s the real part of Kurzweil’s work that I admire. Correct predictions in a book are just credentials that can help the Kurzweils of the world convince our leaders to plan for both the good and the bad of the future. Through his work at Singularity University, and his books and lectures, Kurzweil may convince enough of us to look forward and plan for what may come. I think we should listen. He seems to know what the hell he is talking about.

[photo credit: created from images at Wiki Commons and Accelerating Futures]

Discussion — 18 Responses

  • leonoel49 January 19, 2010 on 10:21 am

    i do not know which one to incline forward, the simplistic approach of Anissimov, or the overwhelming Kurzweil love in this post. i would say some of the “evidence” in the post is rather questionable, like the glasses, which are far from displaying the image directly in your retinas, is more like the old head mounted displays, but smaller and lighter.

    And warfare is far from being dominated by drones, I mean if that is true, Obama is rather stupid sending people to Irak, when clearly drones can get the work done.

    From a certain and very forgiving point of view Kurzweil predictions where accomplished, but I do applaud Anissimov point of view, where he clearly states that trying to give exact dates to certain advancements is no less than being a bit arrogant. Fortunately I am like 40 years older than Kurzweil and will be able to testify whether he achieves or not his so mentioned immortality.

  • James January 19, 2010 on 10:59 am

    My personal opinion is this. Kurzweil has some solid knowledge, he actually is pretty much a genius, but there are two factors which makes him more or less biased: for one, he himself doesn't want to die, so somehow all of these revelations are set to happen within his lifetime (according to his future projections), and from what I know – he also wishes to clone his dad? So that also might distort his logic into more subjective manner. Those are really the 2 things that make me a bit skeptical towards his predictions. I believe that they might be off for maybe 5-20 years and singularity seems a bit far stretched to me.

    I mean, sure it will happen eventually, IF we get to develop those supercomputers everyone is talking about. And that might take a while, a good while. The jump from 2000 to 2010 computer technologies seems like nothing when you are talking about emulating a whole brain/mind, since nor back then, nor now computers are capable of anything like that. Maybe when we discover biological computers or something…but yeah, will take a while probably (though I think it will definitely happen within 21st century.)

  • Vince January 19, 2010 on 7:16 pm

    Well, highways are intelligent in a virtual sense. Traffic is tracked by satellite and utilizes GPS devices. So, to some degree, this helps get you to your destination quicker, and therefore is an assisted form of driving. This isn't far at all from his prediction, even if he were joking about it somewhat.

    Blu-tooth devices are mini computers worn on the ear. So are wireless headphones. They also are starting to market t-shirts that can display images electronically.

    AMOLED screens are fitted onto cell-phones, makes sense that they soon will be fitted onto eyeglasses, and even, contact lenses. There is already see-through AMOLED computer screens, just saw that one today.

    I would say that Ray is pretty much on track with all of his predictions. And you don't have to be a Ray Kurzweil to get criticism, we all have them. :)

  • adsaenz January 19, 2010 on 7:19 pm

    @Leonoel49
    The links in the article are not so much evidence as they are suggestions for further reading. The highlighted technologies are relevant to the discussion, though not necessarily in support of Kurzweil's predictions. I hope that this is apparent in the paragraph after the first numerated list: “As you can tell by my gratuitous use of hyperlinks, many of these technologies are in development if not in commercial use…”
    As for the “Kurzweil Love”…I'm certainly in favor of his conclusions for this debate. And I do admire anyone who is trying to prepare the world for threats of biological terrorism. However, I would characterize myself as a cautious fan when it comes to most of his work and predictions.

  • Bryan "bytehead" Price January 20, 2010 on 1:44 am

    @James: “somehow all of these revelations are set to happen within his lifetime (according to his future projections)”

    According to his future projections, he will live forever anyway. Or, as he puts, live as long as he wants to. Stay alive into the 2030s, and excluding accidents, whatever ails you won't necessarily be cured, but the technology will be there to keep you going until they can cure you. I might add another condition — if you can afford to pay for it.

    I'm not so optimistic. I doubt that I will live long enough to see it, I'll be 72 in 2030 — and I'm younger than Ray. My kids might get that lucky. My grandkids definitely will. The first one is in the oven as I type this. :) IF we don't wipe out the human race or have another dark age. :(

  • Zachary Bottorff January 20, 2010 on 2:17 am

    I have to disagree with Kurzwell “winning” this debate. He clearly stated that his technology predictions would be common place. Obviously the list of seven are not here in every day situations like the cell phone and the Internet are. While he was right these things exist, they were being worked on since the mid 1990's, and that doesn't make it all too surprising to say ten years ago these technologies would be common place.

    Going one by one:
    First: We don't have techno-cloths. Blu-tooth headsets are considered by many to be annoying and snobbish and are desperately being reshaped to be stylish, a heart monitor or hearing aid is meant to be as invisible as possible, and fashion magazines aren't dedicating large sections of themselves to the latest mp3 blasting accessory. We have had wearable electronics, but these fail to hit the computer standard of take input, give output, sort, store, organize, and search. The closest that comes is the iPod, and the ones that do that aren't comfortably worn. Also, I'm throwing out the argument you can strap these things to yourself as I can duct tape a Eee to my arm but that would be a native feature but a hack.

    Second: Speech to Text has been a very slow uptake by those who can use a keyboard. The technology still has problems with people who have accents, speech impediments, or are non-native speakers (which includes the two first groups and a few other problems the programs haven't gotten to fixing completely). This technology is still losing to the keyboard for this very problem. There is a smaller population of computer users who have an issue with the keyboard than voice-to-text.

    Third: We have a device that projects images to the eye. It's used for lazer eye surgery. I bet the technology is coming, but I wont be the first one to put that thing to my face.

    Fourth: I completely disagree with the idea that layered chips are truly three-dimensional. Most of these have a 2D flow pattern, and that just makes it where you have a thick chip with a cooling nightmare and not a real 3D environment. These are gaining ground as they prove themselves to have fixed the kinks inherent in the system, but I would be floored if they covered more than 3% of the active market.

    Fifth: Translating software is the one point I think there is a pro-Kurzwell win. Even though it's horribly flawed, Google Translate and Babblefish are free, online, and readily translate to lots of languages. I again have a problem with him saying “common place” as these translators work as well as a first year language student. “I want to get three of those” will turn into “Three things over there is what I want” is what most of these translators end up making far too often.

    Sixth: Another pro-Kurzwell win. Mainly because he doesn't stay “common place.” And the idea of a flying hand grenade is kind of funny.

    Seventh: Complete fail. GPS doesn't mean I'm not white knuckling cause the out-of-state plates teenager about to side swipe me cause he's trying to blow past semi doing 80mph. What Kurzwell has failed to look at that these system would mean a complete overhaul of every car that gets on the road, which would cost trillions for the United Kingdom alone (Even though I'm an American, I know they use the pound, and I mean in pounds), and probably thrice that per state in the US. If we have Demolition Man style cars it would be after we're forced to give up gasoline and move to pure electric systems.

    Now, I'm unabashedly harsh on Kurzwell, but that's because I feel anyone making predictions shouldn't be overly acknowledged. It's easy to say, “Yeah, this is Sooo going to happen like now!”, but if that was the case then Tucker's super safe auto-mobiles would have outstripped Ford and GM in a short order, NeXT OS would have sank Mac, and Vista wouldn't be the markets current standard for crap (fairly or unfairly, Vista is that standard). Safe cars came decades after the Tucker auto died after car 57 rolled off the line, Apple bought NeXT, and Microsoft is still suffering from the Longhorn-ed debacle.

    Remember that we had the prediction of hover cars and living on Mars from the 60's, but ended up with the internet and cell phones. Progression is not linear, and predictions often make the assumption that they are straight lines from development to deployment. When we get our flying cars, I'll shut up.

  • leonoel49 January 20, 2010 on 3:40 am

    Mhh I would like to add, specially in the glasses part, that both the technologies for creating the kind of glasses you linked to and a direct retina projection is entirely different, one is based on liquid over a screen that gets polarized, the other is based on light transmitted through an even smaller screen to the retinas.

    And on the Self navigating cars, you have to admit that Stanford work on a car that drives itself under a controlled environment with limited navigation capabilities is far from having self driving cars as a common thing on the streets, which is what Kurweil predicts will happen, of course the path is there, but I do not think is safe to say we have it yet, is like saying that machines that understands your movements are here because my work on human recognition and activity by computers have a couple of papers and some preliminary results.

  • OCBandit January 20, 2010 on 6:04 am

    Microvision has been making glasses for the military that “paint” the image directly on the eye ball with a laser. They have been doing this for no less than 10 years. The fact that it hasn't yet become a commodity is probably more a result of component costs.

  • adsaenz January 20, 2010 on 7:11 am

    @Leonoel49
    Yes the Vuzix glasses are linked mainly because they are cool and only slightly because they're relevant to the topic at hand. Mea culpa.

    As you'll notice in the article, I agree with Anissimov that #7 is basically a fail. It could turn around in a few years, but I think the time table is closer to 5-10+ years. Though it is still a likely phenomenon. Check out our latest post on the subject by Steve Wasick:
    http://singularityhub.com/2010/01/20/stanfords-
    Robot cars are getting much better.

    And thanks for all the comments.

  • leonoel49 January 20, 2010 on 2:25 pm

    Sure, no prob, is always nice to have a healthy debate, internet is too full of people with terrible ideas for a discussion. Keep up the good work, I am a big fan of your posts. First time I comment though. :)

  • vessenes January 21, 2010 on 7:59 am

    One thing I was pondering reading Kurzweil's response is the sort of starry-eyed future factor when reading futurists in general. My thought reading Ray's self-defense is that he's technically right that many of those disputed claims are coming, are in development,or are arguably in a pedantic sense happening.

    On the other hand, I was reminded just how normal and human we feel in the face of this significant change in the last 20 years, and that's where I think Ray is being either disingenuous or just missing the point a little.

    Honestly, robot drone armies, personal networks and retina-scanning glasses just sort of sounded cooler when he wrote his predictions. When defending his predictions, he loses this starry-eyed, the-singularity-is-coming feel, and we're suddenly talking nuts and bolts of real technology that you or I can buy (sometimes), plus hard stats on what's been developed or not.

    So, the future, you know, it isn't all it's cracked up to be. But, that's not Ray's fault.

    My AI professor Leslie Kaelbling once said that a tongue-in-cheek definition of AI is whatever computers can't do yet. “Oh, AI is if a computer could read text off a printed page. No, wait, AI is if it can beat you at, like, chess. No, wait, AI is if a computer can recognize what kind of thing you drew and give you the name of it back in written text, and then show you pictures of other things like it…” Once those technologies exist, it's like “Oh, well that's not AI, that's an edge-detection algorithm drilled down to a bunch of eigenvectors and compared to blah-blah-blah.”

    The debate over these predictions feels the same to me. “Well, the singularity is supposed to be unknowable and give us all an incredible spiritual connection with our robot overlords. All I see is my Nexus One offering me more software at a reduced rate..”

    So points to Ray, but he misses the real point: futurists peddle hopes and dreams. When we arrive, say in 2010, we no longer fantasize about it. We've assimilated what we've got, and at the end of it, we're still just people with problems that real people have, even if Touchscreen computers can talk to us, tell us where to go, and only cost $400.

  • Michael Anissimov January 21, 2010 on 9:23 pm

    “How accurately must you predict the future before you can call yourself a futurist? That’s the central question in a bit of online controversy regarding Ray Kurzweil.”

    Absolutely not. I think that Kurzweil is a great futurist. The question here is whether he is accurate in the 90% range or in the 50% range. Because Kurzweil is such a great futurist, it is especially important to analyze whether he is right on-target or slightly off.

  • adsaenz January 22, 2010 on 12:53 am

    @Anissimov
    I would hope we could both agree that we are applying a metric (agreement with real events) to judge Kurzweil's work. I think the application of that metric is essentially asking “Is Kurzweil a good futurist” because I think accuracy in predictions is the biggest part of being a good futurist. If being a futurist to you is more about concern, inspiration, interest, or just general focus on the future then I can see how my opening statement would be inflammatory.
    Still, I think we're talking about the same thing here: Is Kurzweil accurate? How accurate is he? And perhaps more importantly: does his accuracy translate to any real meaningful improvements (in his life, ours, the world's, etc.)
    You and I (in our respective articles/ways) focused a lot on the first two questions. We analyzed his predictions one by one and tried to make judgment calls on what was “common” what was “essentially correct”, etc. Even Kurzweil's own response was focusing on accuracy, and defending his point of view of what constituted “common” ,”correct”, etc.
    After reading over the entire series again, however, I'm inclined to worry that we've all missed addressing the last question in a satisfactory manner. What does Kurzweil's accuracy, his “great futurism” really buy us. I mention briefly it may help leaders prepare against good/bad possible futures. Etc, etc. I think you made the point that Kurzweil's accuracy or lack thereof can help direct us to be more responsible towards predictions in general. I agree that this is an important lesson to be learned, and thus something his work “buys” us.
    But there's probably more to think about here. Has Kurzweil, or any futurist, ever really provided a shift in the way the future unfolded? Economic disasters averted, technological breakthroughs accelerated, peace of mind increased. Those are the sort of impacts that, if we could measure and track down, would be the really important metric for this entire conversation.
    If Kurzweil was only 10% accurate, but that 10% was enough to evoke a major improvement…I hope we could agree that would be a good thing.
    If Kurzweil was 100% accurate, but his predictions had 0 impact…again, I hope we could agree that would be a shame.

    Perhaps we need to think less about accuracy (50 vs 90%), and more about benefits (has anything really changed?).

    Thoughts?

  • Branstrom January 24, 2010 on 1:40 am

    That's not a distinction that Michael made. He was simply saying that 50% accuracy is great, so already at that level Kurzweil would be a great futurist, but 90% is just crazy, superb, and an outrageous claim to make if your serious predictions actually don't hold to a reasonable standard in 9 out of 10 cases.

  • adsaenz January 24, 2010 on 2:28 am

    @Branstorm…
    Yes, I see the importance in being very strict towards those who claim 90% accuracy. At that level they are basically saying you can bank on their predictions. And that would certainly lead to disaster and dangerous speculation in some cases.

    Still, I wonder what constitutes a “reasonable standard”. Kurzweil points out that being off by a few years is totally fine by him. And he has a pretty loose idea of what is “common”. Others want a very strict interpretation. It's all too subjective for me.

    Again, it may be better to evaluate what these predictions have actually accomplished rather than how accurate they were.

  • MakeMake January 25, 2010 on 7:25 am

    “When defending his predictions, he loses this starry-eyed, the-singularity-is-coming feel, and we're suddenly talking nuts and bolts of real technology that you or I can buy (sometimes)”

    The miraculous becomes commonplace through familiarity. I'm sure the Singularity will quickly become “nuts and bolts of real technology” to those living during it.

    If you showed someone a Flip camera or an iPod Touch 25 years ago, they would be very much impressed.

  • MakeMake January 25, 2010 on 7:29 am

    “The question here is whether he is accurate in the 90% range or in the 50% range.”

    Well, since you haven't been able to convincingly argue that it is the latter, why would you pull that number out of your hat?

    Seems like it is the former.

    Excuse, but this is really silly stuff here. 90% accuracy is astounding. It seems churlish for the sake of being churlish to raise objections to such a record.

  • dbharris1964 February 16, 2010 on 2:39 am

    Regarding Kurzweil's so called “fail” on #7, intelligent roads that will allow us to sit back and let our cars drive themselves, please check out this website's own story about Stanford and Audi's latest autonomous driving progress. According to the article, not only do they have cars that drive themselves in real world environments, but now they plan to race an Audi TT up Pike's Peak at full speed. This will require the car to drift and skid under control without flying off the mountain. Check out the video of the car in the desert during a test to see it in action on level ground. If we are at the stage of having sports cars drive themselves up mountains then it is not too much of a stretch to think we will have high end cars driving themselves around at least part time (ex. highway cruising) within 5 years. Self driving cars seem to be coming gradually with incremental changes. First there was cruise control, then adaptive cruise control, then self parking Lexuses and other luxury cars, then lane drift correction (Mercedes) when the driver nods off. The next big feature will have to be some kind of limited highway self driving, like a super cruise control. Think about it, how much brain power does it really take to drive a car down a sparsely populated highway for hours and hours. This new feature would allow people to literally take their hands off the wheel for long periods of time when driving on long highway trips. I think the biggest hurdle will be getting regulatory approval from the DOT to allow this and people's instinctive fear of cars driving out of control crashing into other cars. I'm guessing the opposite will be true, with self driving cars eventually having better driving records over time than human driven cars. The transition will be challenging with some computer driven cars mixing with more erratic human driven cars. Once all cars are computer controlled we will really see almost a complete elimination of “accidents”. During the transition we will probably see human driven cars crashing into computer controlled ones more so than the opposite. Just my two cents.