Monday, January 30, 2017

The Singularity is Near: How Kurzweil's Predictions Are Faring Antropy.co.uk Posted by Paul on January 29, 2017

The Singularity is Near: How Kurzweil's Predictions Are Faring

Antropy.co.uk
Posted by Paul on January 29, 2017
Back in 2007 I was very excited to start reading The Singularity is Near by Ray Kurzweil which had been published 2 years earlier in 2005. As someone who'd always been fascinated by computers and technology since I started programming my first ZX Spectrum as a child, through to my teenage years programming in QBASIC, to my work as a Web Developer, I was well aware of Moore's law and very conscious of the accelerating progress in technology.
As I started to read, it became clear that this was an exceptional book. The clarity of Kurzweil's writing, the level of detail he went in to about each concept he discussed and the content itself were entirely captivating, not least because the predictions were set to happen in the near future and would change the world in to an unrecognisable utopia. In fact, although Kurzweil says early on in the book that his view of the future is neither utopian nor dystopian, the future described is so rosy that the Singularity has also been called "the Geeks' Rapture".
Indeed, if Kurzweil's predictions do come true, they are comparible in their size and implications to the Second Coming of Jesus - the conquering of death and the unification of all consciousness in to one computational substrate that is to become more powerful than anything in the Universe and spreads outward from the Earth at the speed of light ... or faster!
The book quickly became an absolute page-turner for me and as incredulous as it sounded, Kurzweil backed up every prediction with thorough research, logic, and his own impressive credentials - Kurzweil was something of a child prodigy who programmed computers and wrote scientific papers at an early age and became the inventor of OCR (Optical Character Recognition) and the Kurzweil digital pianos and keyboards that bear his name. Kurzweil now works at Google where his job description is "to bring natural language understanding to Google" - a goal not too distant from developing an AI capable of passing the Turing test.
During my time as a software developer, discussions often arose about the future of technology and I would occasionally lend the book to other developers. Unfortunately, at some point the book wasn't returned and I wasn't too fussed as I'd already read it thoroughly. Recently, I decided to buy another copy and re-read it - after all, I still consider it to be one of the best books I've ever read. This time though, 12 years after the initial publication, it would be possible to see how Kurzweil's predictions had done.
Incidentally, Kurzweil himself has released a document describing how his predictions are faring which can be downloaded here: http://www.kurzweilai.net/images/How-My-Predictions-Are-Faring.pdf
Unfortunately though, he doesn't always seem to be intellectually honest with himself:
"PREDICTION
Displays will be built into our eyeglasses and contact lenses and images projected directly onto our retinas.

ACCURACY
Essentially correct
In The Singularity is Near, the wording could be interpreted to mean that it is commonly
used, which is not yet the case, but this is likely to become the case over the next couple
of years."
Well that was 2010, it's now 2017 - where are these eyeglasses and contact lenses? What about Google Glass? Well it may still be being developed, but it certainly isn't in common use. Kurzweil also tends to mark many of his other predictions as "correct", when really they're not, even now in 2017.
So, let's get on to my re-reading of The Singularity is Near, which even for the second time I thoroughly enjoyed and took away a load more fascinating information that I may have missed on the first reading. This time I noted any predictions as I went through that we can now verify and what I found is below.

On pages 11 and 12 we have:

"James Watson, the codiscoverer of DNA, said that in fifty years we will have drugs that will allow us to eat as much as we want without gaining weight.
I replied, 'Fifty years?' We have accomplished this already in mice by blocking the fat insulin receptor gene that controls the storage of fat in the fat cells. Drugs for human use ... are in development now and will be in FDA tests in several years. These will be available in five to ten years, not fifty."
Well, The Singularity is Near was published in 2005, so Kurzweil's deadline for this was 2015. According to the NHS website however, 1,316 gastrc band operations were performed in 2012 and today we still don't have a pill that you can take and then eat what you want. It would be unfair of me not to mention an article on Kurzweil's site: http://www.kurzweilai.net/imaginary-meal-tricks-the-body-into-losing-weight but this has only been tested in mice which Kurzweil even says on p11 had already been achieved. I'm sure Kurzweil would disagree, but if we're being honest, there is still no magic pill available that lets us eat what we want and not gain weight.
Prediction: False

Page 105

"By the end of this decade, computers will disappear as distinct physical objects, with displays built in our eyeglasses, and electronics woven in our clothing, providing full-immersion visual virtual reality. Thus, 'going to a Web site' will mean entering a virtual-reality environment - at least for the visual and auditory senses - where we can directly interact with products and people, both real and simulated. Although simulated people will not be up to human standards - at least not by 2009 - they will be quite satisfactory as sales agents, reservation clerks, and research assistants. Haptic (tactile) interfaces will enable us to touch products and people. It is difficult to identify any lasting advantage of the old brick-and-mortar world that will not ultimately be overcome by the rich interactive interfaces that are soon to come."
Did the above sound anything like your experience of 2009? Because it doesn't seem anything like my experience of 2017. There are a few separate claims here, let's break them down:
  • "Computers will disappear in to clothes and eyeglasses."
    Nope. Computers have become our smart phones though and we're no longer tied to desktops.

    Prediction: False
  • "Going to a website will mean entering virtual reality."
    In 2009, no. In 2017, it is possible that a tiny percentage (and we're talking < 0.0001%) of websites use VR and probably a similar percentage of people have an Oculus, Vive, Google Cardboard, Gear VR or similar VR headset. There is a framework to allow developers to make VR websites called A-Frame but it's definitely not in common use yet.

    Prediction: False
  • "Virtual assistants will be satisfactory sales agents."
    Nope, not in 2009 and not now in 2017. The closest we get is a live chat that is a bot, but as soon as the user realises they're not speaking to a real person, and that is pretty much after the very first question, they would usually rather just use a search or browse an FAQ.

    Prediction: False
  • "Touch interfaces will allow us to touch products and people."
    Sure this has existed in labs for a long time, but do we use this even in 2017? Nope, not even close. Yet another fail for Kurzweil, if we're honest.

    Prediction: False

Page 312

"Computers arriving at the beginning of the next decade will become essentially invisible: woven into our clothing, embedded in our furniture and environment. They will tap into the worldwide mesh (what the World Wide Web will become once all of its linked devices become communicating Web servers, thereby forming vast supercomputers and memory banks) of high-speed communications and computational resources. We'll have very high-bandwidth, wireless communication to the Internet at all times. Displays will be built into our eyeglasses and contact lenses and images projected directly onto our retinas."
  • "Computers in clothing, furniture and the environment."
    Well I have a Christmas jumper with flashing LEDs but I don't think that's what Kurzweil meant, and none of my furniture has a computer in it unless you count the Raspberry Pi on the shelf of my coffee table. As above, this one doesn't seem correct even in 2017.

    Prediction: False
  • "A worldwide mesh will link all devices"
    Many devices do connect to wifi and you can connect to their IP address to access a config page served by their web server, so this seems to be correct.

    Prediction: True
  • "forming vast supercomputers and memory banks"
    In the PDF linked to above, Kurzweil tries to claim he meant the Cloud, but actually I think it's clear from the text that he meant all devices that connect to the Internet will share their resources in terms of memory and processing power. While this probably will happen fairly soon in my opinion and I run the "Folding @ Home" software which donates spare processor time to scientists who are trying to solve cancer and Alzheimer's, and can run on a mobile device, this isn't in common use yet in 2017.

    Prediction: False
  • "We'll have very high-bandwidth, wireless communication to the Internet at all times."
    Correct! Indeed we do, and we use it so much that it's hard to remember that in 2005, we didn't have this.

    Prediction: True
  • "Displays will be built into our eyeglasses and contact lenses and images projected directly onto our retinas."
    Also discussed above, this one seems way off.

    Prediction: False

Still talking about "The 2010 Scenario", Kurzweil continues on page 313:

"These resources will provide high-resolution, full-immersion visual-auditory virtual reality at any time. We will also have augmented reality with displays overlaying the real world to provide real-time guidance and explanations. For example, your retinal display might remind us, 'That's Dr. John Smith, director of the ABC Institute - you last saw him six months ago at the XYZ conference' or, 'That's the Time-Life Building - your meeting is on the tenth floor.'
We'll have real-time translation of foreign languages, essentially subtitles on the world, and access to many forms of online information integrated into our daily activities. Virtual personalities that overlay the real world will help us with information retrieval and our chores and transactions. These virtual assistants won't always wait for questions and directives but will step forward if they see us struggling to find a piece of information. (As we wonder about 'That actress ... who played the princess, or was it the queen ... in that movie with the robot,' our virtual assistant may whisper in our ear or display in our visual field of view: 'Natalie Portman as Queen Amidala in Star Wars, episodes 1, 2, and 3.')"
  • "high-resolution, full-immersion visual-auditory virtual reality at any time"
    Unfortunately this wasn't true in 2010 and isn't true in 2017.

    Prediction: False
  • "We will also have augmented reality with displays overlaying the real world to provide real-time guidance and explanations"
    I'd probably argue that this one is true, although we don't use displays in contact lenses or eyeglasses, it's possible to use one's phone to overlay the world with useful information. An example is an astronomy app that identifies stars and constellations as your point your phone's camera at them.

    Prediction: True
  • "Virtual personalities that overlay the real world will help us ... For example, your retinal display might remind us ..."
    Not even close. We had nothing like this in 2010 and nothing like it now in 2017.

    Prediction: False
  • "As we wonder about 'That actress ... who played the princess ..."
    Well, a personal assistant is not always watching and understanding with the good judgement to know when to interject, but, if we have our phones so configured (and I do), we can talk to our phone and say "Ok Google, which actress played the princess ..." and fairly often a computerised voice will start speaking, not just with the content of the first matching search result, but with a direct answer to the question. Impressive stuff!

    Prediction: True
  • "We'll have real-time translation of foreign languages"
    Speech to text and text to speech are both essentially solved problems. (If you don't believe me, try talking to Google Now on an Android phone.) And language translation has existed for a while. It continues to improve in quality, although bi-lingual friends of mine tell me it's still not great. But it's probably fair to count that this exists.

    Prediction: True

Page 336

"As with all of our other institutions we will ultimately move toward a decentralized educational system in which every person will have ready access to the highest-quality knowledge and instruction. We are now in the early stages of this transformation, but already the advent of the availability of vast knowledge on the Web, useful search engines, high-quality open Web courseware, and increasingly effective computer-assisted instruction are providing widespread and inexpensive access to eduction."
This is essentially correct - the trend toward high quality online courses at affordable costs is continuing and there are some really great resources out there such as:
  • Coursera
  • Udemy
  • edX
  • and of course, one of my favourites, Wikipedia (whose reputation in academia for being inaccurate I don't accept as fair or accurate).
Prediction: True

Page 337

"In the early part of the second decade of this century visual-auditory virtual-reality environments will be full immersion, very high resolution, and very convincing. Most colleges will follow MIT's lead, and students will increasingly attend classes virtually. Vitrual environments will provide high-quality virtual laboratories where experiments can be conducted in chemistry, nuclear physics, or any other scientific field. Students will be able to interact with a virtual Thomas Jefferson or Thomas Edison or even to become a virtual Thomas Jefferson. Classes will be available for all grade levels in many languages. The devices needed to enter these high-quality, high-resolution virtual classrooms will be ubiquitous and affordable even in third world countries. Students at any age, from toddlers to adults, will be able to access the best education in the world at any time and from any place"
  • "early part of the second decade of this century"
    So Kurzweil is talking about 2010 to 2015.
  • "virtual-reality environments will be full immersion, very high resolution, and very convincing"
    If you have a high-end VR headset such as an Oculus Rift, this is indeed correct. At the time of writing they are around £500: https://www.amazon.co.uk/d/PC-Video-Games/Oculus-301-00204-01-Rift/B00ZFOGHRG

    Prediction: True
  • "students will increasingly attend classes virtually"
    This one seems correct, indeed some students do attend some or even all classes virtually.

    Prediction: True
  • "Vitrual environments will provide high-quality virtual laboratories "
    No, unfortunately "virtual" classes tend to just be video calls at this point with IM (Instant Messaging) functionality built-in.

    Prediction: False
  • "Students will be able to interact with a virtual Thomas Jefferson"
    Kurzweil is talking about 2015 but is suggesting there will be virtual avatars capable of some sort of in-character conversation. The Turing test implies that strong AI would be needed for this, and Kurzweil doesn't predict a computer will pass this until 2029, so it's hard to infer what sort of interaction Kurzweil expects of these virtual characters. It's irrelevant anyway seeing as in 2017 we have nothing even close to this in university courses (that I'm aware of, if you know differently please comment below).

    Prediction: False
  • "high-resolution virtual classrooms will be ubiquitous and affordable even in third world countries"
    It doesn't really take any research to state that this one isn't true or anywhere near, even in 2017.

    Prediction: False
  • "at any time and from any place"
    Although Google and Facebook are working on getting Internet to remote corners of the globe, I currently lose my connection when I drive through certain parts of West Sussex, so it's fair to say this one has definitely not yet been achieved.

    Prediction: False

Page 340

"Musicians typically make most of their money with live performances, but that model will also come under attack early in the next decade, when we will have full-immersion virtual reality."
As The Singularity is Near was written in 2005, "early in the next decade" means from 2010 to 2015. Although in 2016 Google released The Bohemian Rhapsody Experience it's probably safe to say that no musician anywhere has lost money because someone wanted a VR band at their wedding or event instead of a live performance.
Prediction: False

Page 341

"By the 2020s, full-immersion virtual reality will be a vast playground of compelling environments and experiences. Initially VR will have certain benefits in terms of enabling communications with others in engaging ways over long distances and featuring a great variety of environments from which to choose. Although the environments will not be completely convincing at first, by the late 2020s they will be indistinguishable from real reality and will involve all of the senses, as well as neurological correlations of our emotions."
  • "By the 2020s"
    It's not really clear what years Kurzweil is talking about here. When he says "By" does he mean before the 2020s? In which case me means in 2019 at the latest? Or does he mean in the 2020s which could be any time before 2029? Either way, it seems these predictions are for the near future so we cannot definitively say true or false but we can see how close we are.
  • "full-immersion virtual reality will be a vast playground of compelling environments and experiences"
    There is already some good VR content in 2017 and the experience is pretty good. There could definitely be more content though.
  • "enabling communications with others in engaging ways over long distances "
    Currently it doesn't look like VR will be used for video call type communication, it will be more about exploring environments, albeit perhaps collaboratively. 
  • "by the late 2020s they will be indistinguishable from real reality"
    This isn't at all unbelievable considering the late 2020s are over 10 years away and some computer games are already looking very close to reality. It's often hard to know if a video is from a computer game or not. Making games more realistic is essentially just a case of more graphics processing power, so in 10+ years it seems very likely that games will be photoreal. Connecting VR to an existing game is easy, so it doesn't seem there will be any difficulty with this prediction.
  • "will involve all of the senses"
    This will be harder to achieve because it involves very advanced hardware or nanobots that connect directly to neurons. Perhaps we will develop those, but 10 years seems too short for this one, although this is of course my own personal feeling and speculation.
  • "neurological correlations of our emotions"
    By this I assume Kurzweil means that either nanobots in our brain or some other technology will directly cause emotional experiences. I think Kurzweil means that it would be like watching a film and being put in to appropriate moods and being given certain emotions by some unknown technology. Personally I find a good musical score is more than capable of achieving this but I'm not sure being forced to feel a certain way is desirable even if it is technologically achievable as in the case of people undergoing brain surgery while awake who react with certain moods when electrodes are placed in specific places.

Page 424

"For example, before this decade is over, devices the size of dust particles will be able to carry out reconnaissance missions."
Before this decade is over here seems to mean by 2009. Kurzweil is an advisor to the US Military and mentions in the book that he isn't allowed to disclose everything he knows. So it may be the case that dust particles are already carrying out missions for the US Military but I can't help thinking such a technology would be in use in business as well if it existed and I've not seen any news about such a thing. For it to work there would have to be nanobots that could move around an environment on their own power and transmit a signal back wirelessly. I'm just not sure we're there yet with that one.
Prediction: False (unless the US Military has these and they're classified which I find implausible)

Page 473

"In the second decade of this century, we will routinely be interactive with virtual humans that, although not yet Turing-test capable, will have sufficient natural language understanding to act as our personal assistants for a wide range of tasks."
I think it's probably fair to say that OK Google can perform this role. Once enabled on your Android device, you can open your phone and say things like "OK Google ..."
  • "Set an alarm for x"
  • "How old is x?"
  • "Who invented x?"
  • "Navigate to x"
  • "Call x"
  • "What time is it in x?"
  • "Do I need a jacket tomorrow?"
  • "What is the current share price for x?"
  • "What is x GBP in USD?"
  • "Calculate x times y squared?"
  • "Define x"
  • "Take a picture"
  • "Play {song name}"
And these are not gimmicks. My experience has been that the voice recognition is incredible, even in incredibly noisy environments. Often while reading, I'll ask Google to define a word. Almost always, it will get the exact word, even considering there is no context and if I'm looking it up it's probably a word that I don't know how to pronounce. I'm certain it can do this better than a person would, especially with very obscure words.
Prediction: True

So, how near is the Singularity?

I am still a Singularitarian ("Ich bin ein Singularitarian" as it says in the book!) which means I still think that, after looking at as much evidence as I can in books and the real world, it's likely that the exponential progress of technology is a continuation of biological evolution and will continue to accelerate until an intelligence emerges that is able to improve itself and at that point, the pace of progress will accelerate beyond recognition.
So to me, the real question is about the timescales involved. Let's have a look at how Kurzweil did in the number of predictions he's got right so far:
  • 25 Predictions
  • 9 True
  • 16 False
This is actually quite impressive! Although 9 out of 25 is only a 36% accuracy rate, I still remember when reading The Singularity is Near for the first time that almost all of the predictions seemed wildly optimistic and sort of crazy. It seemed a bit unlikely that it would be possible to have a high speed internet connection from a touchscreen super-computer everyone has in their pocket that can also act as a personal assistant that you can speak to in natural language and it will usually understand and respond appropriately, albeit unable to have a full conversation at this point.
Throughout the book, Kurzweil includes graphs of processing power and computer memory for a given dollar value. This growth, related to Moore's law, Kurzweil calls The Law of Accelerating Returns. The graphs have remained spot-on. In fact there is a prediction listed on the Wikipedia page of Kurzweil's predictions that 10TB of computer memory will cost $1000 or less in 2018 - it's been removed now but the Wikipedia page previously linked to a 10TB drive on Amazon from 2016 which was around $600.
While Kurzweil's predictions for computation and memory capacity are extremely accurate, sometimes his predictions for the technologies that will be in use are wide of the mark. It might be that these technologies are indeed easily technologically possible but they just aren't that desirable. Kurzweil talks at length about computers being woven in to our clothing but doesn't mention the smartphone. I can't really see why we'd want a computer in our clothing (or indeed our washing machines if we're not careful) but it could be said that he was fundamentally right that we would have a powerful, connected computer with us at all times.
Many of Kurzweil's predictions relate to Virtual Reality, and while Oculus enjoyed a massive following and was bought by Facebook, and many other commercially successful VR headsets are available, their high price, awkward usage characteristics (wires, setup, bulk etc.) and the fact that they mostly only appeal to geeky gamers has meant that adoption is currently very low. It's even possible that VR is a fad like 3D TV which never really takes off however hard the manufacturers try, or it could be that Facebook manage to find a social, people-centric use for VR that improves the Facebook experience so much that it suddenly takes off.
Interestingly, Kurzweil mentions fairly early on in the book that self-driving cars are a long way off and doesn't even bother to give a date. He must have considered it a pretty formidable challenge but Google, Tesla, NVidia and others seem to be very close now to having viable self-driving cars on the roads, so while some predictions are way behind schedule, others are way ahead of schedule.
It seems that Kurzweil's mistakes then are more related to a mismatch between what is technically possible and what is desirable. No one wants a computer in their jumper but everyone wants the latest new smartphone. There isn't a massive demand for VR outside the gaming community but there is for high capacity hard drives. So it's more likely to be a combination of what's possible and what consumers want that shapes the technologies that become ubiquitous over the coming decades.
The biggest sections of the book are about the three coming revolutions in GNR (Genetics, Nanotechnology and Robotics/AI) and I hope to revisit the book in another 10 years to see how these things have progressed.
In terms of a date for the Singularity itself, if Kurzweil's estimate of the computational capacity of the human brain is accurate, and his logic that increasing brain scan resolutions will allow us to reverse engineer the entire human brain and model it in a computer, there is no reason to doubt his date of 2029 - other than how incredulous it sounds! The Singularity may indeed by Near.

1 comment:

  1. Do the graphic training institutes provide placement services? Subhe offers 8 months program in graphic design classes that will give you the broad experience of graphic design. Learn more aspects of Digital, Graphics, UI/UX & Web Design from the best online graphic design courses. Apply Today.

    ReplyDelete