Interviews: Ask Ray Kurzweil About the Future of Mankind and Technology

The recipient of nineteen honorary doctorates, and honors from three U.S. presidents, Ray Kurzweil's accolades are almost too many to list. A prolific inventor, Kurzweil created the first CCD flatbed scanner, the first omni-font optical character recognition, the first print-to-speech reading machine for the blind, the first text-to-speech synthesizer, and the first music synthesizer capable of recreating the grand piano and other orchestral instruments. His book, The Singularity Is Near, was a New York Times best seller. and is considered one of the best books about futurism and transhumanism ever written. Mr. Kurzweil was hired by Google in December as Director of Engineering to "work on new projects involving machine learning and language processing." He has agreed to take a short break from creating and predicting the future in order to answer your questions. As usual, you're invited to ask as many questions as you'd like, but please divide them, one question per post.

Comments

Your Predictions

[eldavojohn]

Editors of Wikipedia have taken enough care to meticulously log your predictions. Are there any that you regret making? Are there any that you think people overlook because now it's painfully obvious but wasn't at the time?

Your Countdown to the Singularity

[eldavojohn]

I have seen the graphic showing your countdown to the singularity and something I've always wondered is how you picked these events and what makes the significant? For example, your list seems to be made of things that would prolong our existence but entries like "Human ancestors walk upright" and "art, early cities" are confusing in that I don't understand how they can be marked as epic achievements. Are you saying that if we had never learned to walk upright we would not have developed intelligence? Are you saying that early cities were somehow superior to ant colonies? Didn't they help spread disease and cause sanitation problems? Can you convince me that this list isn't just arbitrary things that fit into a line?

Re:Your Countdown to the Singularity

[Colonel Korn]

I have seen the graphic showing your countdown to the singularity and something I've always wondered is how you picked these events and what makes the significant? For example, your list seems to be made of things that would prolong our existence but entries like "Human ancestors walk upright" and "art, early cities" are confusing in that I don't understand how they can be marked as epic achievements. Are you saying that if we had never learned to walk upright we would not have developed intelligence? Are you saying that early cities were somehow superior to ant colonies? Didn't they help spread disease and cause sanitation problems? Can you convince me that this list isn't just arbitrary things that fit into a line?

They were chosen because they create a line on a log-log plot.

Re:Your Countdown to the Singularity

[Amorymeltzer]

entries like "Human ancestors walk upright" and "art, early cities" are confusing in that I don't understand how they can be marked as epic achievements. Are you saying that if we had never learned to walk upright we would not have developed intelligence?

Walking upright allowed us to freely use our hands. Everything since was really put on hold until that could happen. Our hands are dextrous and capable to a degree of finery not seen elsewhere. That event was huge. Art is a sign of a creative intelligent process, I don't understand why you don't see that as something "epic." Art is an enormous part of what we call "culture" which is an enormous part of what sets us apart from our evolutionary cousins.

Are you saying that early cities were somehow superior to ant colonies? Didn't they help spread disease and cause sanitation problems?

Yes, if only because humans are superior (at least as far as intelligence goes) to ants. There's no point in grouping together the smartest viruses for a few thousand years and hoping to get a laptop, nor would you advise a company to hire only those with room-temperature IQs. Diseases and sanitation problems, sure, but the population has moved past those issues by an order of magnitude or three, at least in many places.

Re:Your Countdown to the Singularity

[Amorymeltzer]

That you think so lowly of Neanderthals and cavemen is just proof of how critical art has been. We've all got some Neaderthal admixture in our history, they weren't truly so different from Homo sapiens (sapiens) cavemen. Humans largely indistinct from us have been around for tens of thousands of years, but it was the development of things like art, language, and civilizations that has allowed us to persist. Art can be used to tell stories and pass on information, and indeed was in the largely illiterate history of our ancestors. Art conveys ideas, and ideas are how you get Tesla and plastics. Don't tell me Isaac Asimov or Arthur C. Clarke or Da Vinci weren't pivotal to human intelligence because they created "art."

Re:Your Countdown to the Singularity

[Genda]

Actually the whole point of cities demanded that human beings address the very things you mention and more, inventing culture, laws, technology, agriculture, common language, modern commerce and architecture. Cities are the birthplaces of social evolution.

For those of us old enough to remember...

[benjfowler]

Is Ray Kurzweil the new Jon Katz?

What kind of inmortality?

[javilon]

What do you think will come first, immortality through repair technology like SENS, or immortality through mind uploading?

Have Human Enhancement Technologies Slowed Down?

[eldavojohn]

I assume that the occurrence of Human Enhancement Technologies (HETs) needs to accelerate for us to hit the singularity in 2045 as you predict. While we cover a lot of them on Slashdot, they either feel like vaporware or just a small improvement on an existing HET. Of the existing technologies in actual use they all seem a decade or more old. So where is the acceleration of HETs and their proliferation? Why am I not seeing more normal humans using HETs or at least more original HET options arising? Can you explain what I'm missing?

Is Google's goal a singularity?

[Psyborgue]

Is creating a singularity a goal (immediate or long term) at Google?

Re:Is Google's goal a singularity?

[the gnat]

The real question is should Superintellignece be developed first by the private sector (Google) or by the public sector (Government)? Who should get it first and why?

The one without guns and nukes, of course. I'm not a reflexive defender of the private sector versus the government (hell, I'm employed by the government), but I'm hardly so naive as to think that just because the government creates something, it's "mine", any more than it would be if Google created it. If the US government invents superintelligence it will probably just use it to spy on American citizens.

Extraneous human population

[pkbarbiedoll]

As technology advances particularly with regard to robotics and AI, we're going to find that a large segment of the human population simply is not needed anymore. In today's political environment I'm simply not seeing the global community embracing strict population control as well as socialism in providing for those who no longer have jobs and are simply using up resources without providing anything in return.

What do you recommend be done with these billions of people in the coming decades?

Re:Extraneous human population

[spire3661]

I fucking hate this. What do you mean 'not needed anymore'. My job is not my purpose in life. I exist to exist, not to work until i die.

Re:Extraneous human population

[bill_mcgonigle]

What do you recommend be done with these billions of people in the coming decades?

Blech. The question assumes that anybody can make an informed decision about what to tell billions of people to do.

Personally, I'm looking forward to ten billion people who have all the food, energy, materials, and information they desire, and can't even begin to imagine the beautiful things that will come of it (other than a gradual reduction of that ten billion over time). The music we'll hear, the stories we'll read, the advances in science and engineering that I'll see some day, and the amazing amazings I can't even guess at!

Re:Extraneous human population

[Golddess]

What do you mean 'not needed anymore'.

I believe pkbarbiedoll means that you've been fired and a robot has taken your job. Sure, your job is not your purpose in life, but you may now find it difficult to put food on your table and keep a roof over your head. So what, if anything, should society do for such displaced individuals? Are current unemployment benefits sufficient, or should we plan for some huge influx of people? What, if anything, needs to be done to avoid a new Luddite movement? Sure, we survived the first, and I am not claiming that we would not survive another. But would society have been better off if something* had been done to avoid the Luddite movement entirely? Should we just let a new Luddite movement run its course?

*Something other than stopping/reversing technological process.

Re:Extraneous human population

[bzipitidoo]

No kidding. If the GP knows the meaning of life, I should like to hear it. People and life never were "needed". Like any inanimate object, the Earth would not miss life at all were it absent.

It's up to us to decide how we want to live. In many ways it isn't even up to us. Like all animals, we have instincts that dictate a great deal of our behavior, honed by millions of years of evolution. Malthus worried that unrestrained greed would lead us to populate as much as possible even though it would have to end in a "tragedy of the commons" catastrophic collapse. But actually, life dealt with this problem long before we were around. If resources are scarce, animals don't breed. Quite a few animals, such as kangaroos, even abort pregnancies if times turn hard. Takes a lot of energy to produce and raise offspring, and so animals evolved to save themselves the trouble and not to do it if failure looks highly likely. Since it is the female that bears most of the costs, she has evolved to make the call. Today, in nations where women have a choice, women are having fewer children than ever. Where women don't have a choice, we have overpopulation, ecological trouble and war. The number one reason people go to war is that it may be better than doing nothing and starving. This is a problem with an easy solution, if only we are willing. Of course, unanticipated disasters can still lead to tragedy, but that's life. We may not ever able to predict the future well enough to avoid every famine.

We have an easy answer to that problem, the only question there is whether we're willing. If we are, then we can focus on other matters. Where do we want to go? What kind of life do we want to live? Why shouldn't all of us who want it be able to have a life of fun? Spend our days skiing, rock climbing, white water rafting, partying, and having sex without undesired pregnancies, diseases, and love triangles that turn ugly? When bored of all that, shift to different pleasures, such as hacking and creating art? The very people who seem to object most to that kind of existence, out of some notion that that's the deadly sin known as Sloth with a bit of Lust and Gluttony thrown in, a bunch of immoral hedonism that will lead to divine punishment, are the same who seem unwilling to face one of the big challenges of our generation, that of Climate Change.

On patents

[godrik]

You invented lots of things that proved to be very useful to a wide range of people and industries. While the patent war is going stronger than ever, do you believe that you could have succesfully develop so many ideas in the current legal context?

P = NP?

[Karganeth]

I am certain that the algorithms being used today will never result with human level AI. If we want to design strong AI we need to first prove P = NP. This is the missing key to solving our AI problems - it's why there's a huge difference between expectations and the reality of AI. I will spend much of my life trying to prove P = NP, no matter how many tell me it's completely futile. Do you think P = NP?

Re:P = NP?

[Karganeth]

If P = NP, then the world would be a profoundly different place than we usually assume it to be. There would be no special value in "creative leaps," no fundamental gap between solving a problem and recognizing the solution once it's found. Everyone who could appreciate a symphony would be Mozart; everyone who could follow a step-by-step argument would be Gauss... — Scott Aaronson, MIT

I think that this is a more accurate description of how creativity and intelligence works. People claim they "invented" or "created" a new machine or song, but in reality they just copied it from somewhere else in nature (with or without realizing it).

Re:P = NP?

[VortexCortex]

I am certain that the algorithms being used today will never result with human level AI. If we want to design strong AI we need to first prove P = NP.

I am certain that you are ignorant about much of the field of machine intelligence.

Must you prove P = NP to become sentient? Think man! I can train a small neural network to do OCR, or recognize shapes or colors, with more complexity it can do more things. I can train a Dog's neural network to do many tricks, the same for Parrots. Why would a digital neural network be any less capable of sentience than your own mind? I do not need advanced mathematics knowledge beyond our current understanding to create a framework for neuron simulation and scale it up to human levels of complexity, then genetically program it via selection pressure.

The only thing that we need is bigger and faster machines to run the scalable neural network simulations, and time to train it. Ergo, it's only a matter of time. If my OCR neural network doesn't have to be self aware to do OCR, and your brain doesn't have to know whether P = NP to be sentient, then neither does Strong AI. The intelligence of my machine's neural networks is not artificial -- It's as real as that of any living neural network. The network is artificial, but the intelligence is real! Intelligence isn't invented, it's a phenomena that emerges from any sufficiently complex interaction.

Any sufficiently complex interaction is indistinguishable from a sentient being because that's what a sentient being is. What's designed in my MI systems is merely a scalable framework for efficient interaction and reflection; Datasets then dictate the flow of information in that framework, and they are EVOLVED to produce the results desired -- Evolution through Selection Pressure is something we CAN prove exists, and it's all we need.

To truly UNDERSTAND how strong AI thinks might require a proof that P = NP, but for the strong AI to actually exist does not require this knowledge. In other words: I couldn't tell you exactly how my OCR program learns to recognize the letter Z at any angle, I just know that it has evolved the encoding of such an understanding because that's what I trained it to do -- I don't know how it works exactly, it just does (to a sufficient degree). I can explain/document one instance of the series of changes that resulted in that capability, but I can not completely explain the process precisely. I have tried decoding the logic pathways of advanced cyclic neural networks and become mired in a tangle too deep for anyone to fully understand, especially considering that the same knowledge can be learned by a myriad of different AI frameworks, where my documented process diverges sharply from the process employed in the others. Pattern recognition can be learned by info. networks, and genetic programming can be used to produce better pattern recognition, and that's all the understanding we need to create strong AI -- Syntax is merely an application of pattern recognition. Your abliity to read the words of this sentence without sounding them out is an application of compressed pattern recognition -- esp. the second misspelled word. Mathematics and Science are merely an application of compression in a cybernetic system: A preliminary processing of data produces a pattern that is recognized to indicate that further processing will result in an expected result (previously processed fully).

Once you know a bit of information theory it's ridiculous to be so Chauvinistic about your own Sentience -- You're not special! I suspect you've been reading too much into Asimov and the whole "laws of robotics", and other such nonsense. My machine intelligences are not based on expert systems, and if they become complex enough to develop self preservation they may very well try to harm a human that's attempting to harm them -- As will a Dog, or Parrot, or Bear, or any brained creature with sufficiently complex interactions going on in their heads

Immortality

[Amorymeltzer]

It's been said that the first person to live forever has already been born. In what sense is this conceivably true? How would such a medical/technological advance affect society, and how on earth could we avoid something catastrophic from occurring? (If you've ever read the Red Mars trilogy, I think of what happened when the longevity treatment was introduced)

let me fix this for you

[stenvar]

That should be:

A prolific self-promoter, Kurzweil claims to have created the first CCD flatbed scanner, the first omni-font optical character recognition, the first print-to-speech reading machine for the blind, the first text-to-speech synthesizer, and the first music synthesizer capable of recreating the grand piano and other orchestral instruments

Most of these claims are actually rather dubious.

Re:let me fix this for you

[Anonymous Coward]

Let's see. The first synthesizer able to play sounds of natural instruments... can it be Mellotron (1962), which has sets of magnetic tapes for different sounds? Kurzweil was 14 then. Text--to--speech electronic devices? Some are from early 60s as well. Omni--font OCR? From Wikipedia: Kurzweil is often credited with inventing omnifont OCR, but it was in use by companies, including CompuScan, in the late 1960s and 1970s. Print--to--speech? From Wikipedia: In 1949, RCA engineers worked on the first primitive computer-type OCR to help blind people for the US Veterans Administration, but instead of converting the printed characters to machine language, their device converted it to machine language and then spoke the letters: an early text-to-speech technology..

Furthest future entry in your personal calendar?

[ewg]

What's the furthest future entry in your personal calendar?

The Premise of Conflict in All of Earth's History

[eldavojohn]

Something that bothers me about the singularity is the complete removal of conflict. Okay, we've cheated death eternally, we are merged with machines, nationality is a distant memory and Earth is completely terraformed to be computing space for our vast artificial intelligences. There will no longer be man vs man or man vs environment. Where is the conflict? What causes us to strive for anything? It sounds like a veritable utopia and I should just kick back and let it happen. How will progress be made without conflict?

D-Wave quantum computer

[Simon321]

What do you think about the D-Wave quantum computer? Do you think it will be able to display a 'quantum speedup' over conventional computers? If it does, what will the consequences be for your singularity roadmap?

Re:The Premise of Conflict in All of Earth's Histo

[kruach aum]

Out of interest? While I won't deny that conflict is a major innovator and force of progress, I'm pretty sure Einstein didn't come up with special relativity because of how much he wanted to embarrass Newton. Rather, he was interested in what simultaneity meant, and then started to think. I'm not sure why things would be different after the Singularity. It could even be argued that conflict is an inhibitor of progress in some cases: Darwin didn't publish his work on evolution for twenty years because of the conflicts he foresaw. It was only his discoverer's pride that stopped Alfred Russell Wallace from getting the scoop.

I'll go for it

[smittyoneeach]

What if what seems to be a general technology-driven implosion into a Single World State goes a completely different direction, due to some 'unexpected' encounter with the fundamental irrationality of humans? In short, are there hidden assumptions about the rationality of human nature within this 'Singularity' notion?

Why progress?

[vlm]

Progress... why? You express it as a function of time. Why not a function of "energy controlled"? So when the EROEI of crude oil and/or coal drops below critical value then progress stops and regress begins, right?

The third Industrial Revolution impact

[Sique]

As mentioned on Slashdot before, Bob Gordon argues that there have been three Industrial Revolutions, the Steam Revolution in the late 18th century, the Electrical and Car Revolution at the turn of the 20th century and the Data Revolution since the 1950ies. Differently than the first two, which yielded immense productivity and wealth increases, the Data Revolution is not living up to its promises yet, though we have many of its aspects already in place, data processing power is at everyone's[*] disposal, a world wide communication network lets you reach a big part of the world population[*] for nearly zero cost, a tremendous source of information is readily available to everyone[*].
[*] everyone either living in the Northwestern hemisphere or being wealthy and influencial enough.
1) Do you agree with Bob Gordon's notion?
2a) If yes, why is that, and will it change in the near future?
2b) If no, where do you see the great increase in productivity and wealth, Bob Gordon is missing?

Re:The third Industrial Revolution impact

[T.E.D.]

I can answer this one. Bob Gordon is looking at the wrong things. The real "revolutions" in history have all been information revolutions. Some poeple get too mesmerized by the effects of them to look for the common causes. You have language, then writing, then printing, then electronic information. Each one allowed members of the society that used it access to orders of magnitude more information than previous ones, essentially putting them in a whole different class.

The implication of this is that if you really want to advance society, find it a more efficient way to copy, store, and process information.

Re:Alien Singularities

[cuncator]

Probably the best quote I've heard that addresses this is from Calvin and Hobbes: “Sometimes I think the surest sign that intelligent life exists elsewhere in the universe is that none of it has tried to contact us.”

Dangers of Artificial Intelligence

[ahbond]

Is AI more dangerous than nuclear weapons? I'd like to hear your thoughts on how/if this technology should be controlled / monitored.

Re:Have Human Enhancement Technologies Slowed Down

[ColdWetDog]

Just as a sidelight to your comment that might point towards the answer (and I'm no particular fan of the Singularity thingy) - we ARE seeing a good bit of HET in the guise of 'cheating' in the Olympics. Most of these are pretty crude chemical approaches although gene manipulation therapy is probably the next big step.

These sorts of things are very under the radar, clandestine if not outright illegal.

You may need to see something like William Gibson's Chiba City start along before we see much progress.

Hell, in the US, we're still stuck on putting marijuana in the same category as heroin. You can't expect all that much from this sort of society.

Re:The singularity has been "near" for decades.

[ecloud]

Siri? Google Glass and the apps that run on it? Google Voice turning your voicemails into emails as fast as you receive them? Turn-by-turn directions developed independently by several companies? This stuff used to be called AI.