An Interview with Hugo de Garis
Currently a professor at Xiamen Universty, where he teaches theoretical physics and computer science, Hugo de Garis is legendary in AI and futurist circles. He’s been working on neural networks and “artificial brains” since the 1990s. de Garis is also an advocate of Femtotechnology — the notion that we could operate technologically at a level a million times smaller than nanotech. In 2005, de Garis published a controversial book called Artilect Wars that predicted an upcoming “gigadeath” war between Terrans (those who want to remain human) and “Cosmists” (those who want to unite with AIs and become posthuman and post-terrestrial).
H+: You were involved in a project called ATR’s Human Information Processing Research Laboratories (ATR-HIP) that aimed to create a billion neuron artificial brain by the year 2001 and I’m assuming you didn’t quite make it. But what did you learn from that project?
Hugo de Garis: I got close. The aim was to build a machine that would evolve neural net modules at electronic speeds. It did that. I could evolve a neural net module in seconds. Unfortunately, the lab that took over the work, Starlab, in Brussels, went bankrupt in 2001 with the dotcom crash, and the whole project died due to my hardware designer wiping his hands of the whole affair. (He lost $300,000 of his own money over the bankruptcy. He was not paid for his machine sold to the lab).
Nowadays, with NVidia’s Tesla and Fermi machines, available at only $10,000 each, the old dream of nearly a decade ago is again practical, thanks to Moore’s Law. So the Chinese lab I was at (I’m now retiring), taken over by Ben Goertzel, will evolve large numbers of neural net modules to build artificial brains. I’ll be describing all this in a book that Ben and I will write this year called Artificial Brains. Ben and I have just finished a special issue on the theme of Artificial Brains for the Neurocomputing journal, the first of its kind in the world.
H+: But… why are you retiring?!!
HdG: I’ve been building artificial brains (mainly evolving neural net modules at electronic speeds) for 20 years. That’s a long time. During that time, my interests shifted. My concentration increased (I’m not sure if that is due to the Omega-3 fish oil tablets I take each day, or a result of the discipline training of being a professor who simply has to master new material so I can teach it the next day, or just the maturity of aging… becoming more scholarly.) The net result is that I’m now much more interested in PhD level pure math and math physics than programming software to design artificial brains. Since I’m living in China, my US savings go 7 times further, so I can retire now and then, later, live quite well from my father’s inheritance. I’m now intensively studying pure math and math physics, staying up all night, doing what I want to do. The freedom is intoxicating. I will visit my old lab once a week to stay in touch with my friend and successor, Ben Goertzel, and have yet to finish a book on artificial brains.
H+: Please tell us a bit about topological quantum computing (QC) and femtotechnology. When do you think it will become a reality and how powerful will it be?
HdG: QC is a revolution in computer science (CS) and CS teaching. It’s been known for a quarter century that quantum computers are exponentially more powerful than classical computers, so when they come, the 21st century will be revolutionized. The problem with today’s QCs is that they are so vulnerable to local disturbance. If you store a bit on the spin of an electron, for example, then it is highly vulnerable to local noise — that can change the state and hence the information. Topological quantum computing (TQC) promises to store information in topological quantum fields, i.e. spread out over a field, so it is robust against local noise, thus allowing QCs to scale up. In 2005, experimental physicists found the first solid-state quasi particles (“anyons”) that have these topological quantum field properties. The math explaining how to use these anyons to perform TQC has been known since 2002. The physicists are now in a race to find anyons that can be used for universal computation. That should happen in a year or less. They say they are very close.
Given the critical impact that TQC will have on society, it will be essential that its principles be taught, but that will revolutionize the teaching of computer science, since to understand TQC requires a knowledge of graduate (PhD level) pure math and mathematical physics, e.g. on the math side, algebraic topology, cohomology, quantum groups, knot theory, representation theory, Lie theory, etc. On the physics side, quantum field theory, topological quantum field theory, conformal field theory, condensed matter field theory, Fractional Quantum Hall Effect, etc.
So, the curriculum of CS departments around the world will be totally shaken up. CS will become much more mathematical physics-based. I talk about the “TQC CS teaching shock wave.” I’m contracted after finishing the Artificial Brains book to write another one on TQC. I’ll write it in 2011-2012.
As for femtotech? I remember asking Eric Drexler in the early 90s, when I was visiting MIT, whether he thought femtotech (femtometer scale engineering) would be possible. He pooh-poohed the idea, which struck me as ironic, because he did to me what his early critics did to him on nanotech. One of my major goals is to study pure math and mathematical physics very hard during my retirement and see if I can find physical phenomena that could be used as the basis for a femtometer scale technology (computing, engineering etc). Since a femtometer is about the scale of a quark, I will need a lot of theoretical physics knowledge. I plan to write a book about my ideas a few years from now, if I’m not beaten at the post.
Femtotech would be to nanotech what nanotech is to millitech, i.e. the scale would be a million times smaller, hence the computing capacity (assuming heatless) would be a trillion trillion times greater, i.e. a million cubed for the density of components, times a million times faster, because the component distances are a million times smaller.
H+: Has their been any evidence for people being able to do stuff on that femtotechnical scale. And don’t quarks behave incomprehensibly… isn’t it pretty much pure dadaism down there?
HdG: I’ve been fascinated by the question of whether a femtotech can exist, for over 20 years. It’s the obvious successor to nanotech. The last time I googled the term, all I got were other people asking the same question, but no one had any answers. So once I’m fluent in PhD level pure math and math physics, I will go on a systematic sniffing around to try to find some answers, i.e. potential phenomena in physics that might serve as the basis for a femtometer scale technology (femtotech), and then write a book about it. I am strongly encouraged by Ben Goertzel to do this. If successful, I would become the Eric Drexler of femtotech.
H+: Your book, Artilect War, describes an upcoming war between “Cosmists” and Terrans… and some leading scientific thinkers have criticized this vision. Do you still believe this is likely, and have any particular counterarguments struck you as particularly worthy?
HdG: I guess my main critic over the years has been Ray Kurzweil, who says that if it came to a war between the Cosmists (who want to build artilects— artificial intellects equal to godlike massively intelligent machines) and the Terrans, who are opposed, it would be like “the US Army fighting the Amish.” I agree, if the Terrans sit back and let the cyborgs and artilects come into being. But the Terrans are not fools. They will be horrified by the cyborgs all around them and know that they cannot wait too long if they are to keep a chance of winning a war against the Cosmists/Cyborgs/Artilects. They will have to strike first. But the Cosmists will anticipate this and be ready, and since all this is likely to happen later this century, with later 21st century weaponry, the scale of the killing in the most passionate war ever will be in the billions — what I call “gigadeath.”
Many surveys on this question (are you Cosmist or Terran?) split people down the middle. The rise of the cyborgs and artilects will profoundly disrupt humanity, and millions will oppose it. The Terrans will fear their fate at the hands of the artilects and the cyborgs. (A cyborg is virtually indistinguishable from an artilect, given that the computing capacity of a grain of sand is trillions of times greater than the human brain). As the gap between human level intelligence and home robot level intelligence (based on artificial brain technology) closes, the “species dominance debate” will heat up. It’s starting to heat up already amongst the techies and AI types. The media are starting to get interested, e.g. History Channel will film me for a fifth of a two hour special on the theme of “futurists” in September. Murphy’s Law (if something can go wrong, it will) will have enormous scope with the rise of the cyborgs and artilects. The Terrans will live in fear, which will motivate them powerfully. In the 20th century, wars were about the survival of a country. This century, an artilect war would be about the survival of a species, i.e. us, human beings. Given that the stake has never been higher, the passion level in the species dominance debate will be extreme. I’m glad I’m alive now, so that I do not have to live through the horror of seeing billions of people die, but my grandson, one year old, will see it, and probably be destroyed by it. It’s a very gloomy message, but realistic, I think. The two ideologies are very powerful. The Cosmists will want to build gods… that the artilects could become. The Terrans want humanity to survive and will be prepared to exterminate a few million Cosmists for the sake of the survival of the billions of humans (or early cyborgs). These two ideologies will dominate our global politics this century. They will color the age. They will replace the issue that dominated 19th and 20th centuries politics, i.e. “who should own capital?” The 21st century will be dominated by the question “Do we build gods… or our potential exterminators?”
My second book, Multis and Monos, is about the rise of a global state. In 40 years, there won’t be any dictatorships left, and the internet will be a trillion times faster, resulting in the creation of a global language, a global culture and a global state, so no more wars, no more arms trade, no poverty, all very optimistic. But even a global state will not be able to stop the artilect war. The species dominance issue is too powerful, too passionate — there’s too much at stake.
H+: So what’s your take on Ben Goertzel’s Cosmism, as expressed in A Cosmist Manifesto? [editor’s note: look out for a discussion here on H+ magazine this coming Thursday, 9/2/10, about A Cosmist Manifesto, now available as a print book, published by Humanity+Press.]
HdG: Ben and I have essentially the same vision, i.e. that it’s the destiny of humanity to serve as the stepping-stone towards the creation of artilects. Where we differ is on the political front. I don’t share his optimism that the rise of the artilects will be peaceful. I think it will be extremely violent — an artilect war, killing billions of people. I wouldn’t be surprised with the trillion trillion stars out there that this transition from biological to artilectual that humanity is about to undergo is a cosmic phenomenon that has already occurred a zillion times in our universe. It’s probably just part of natural cosmic evolution. It’s not for nothing that I have labeled the Cosmists… Cosmists.
H+: So I take it that your artilects will not be developed by reverse engineering the human brain ala Kurzweil. Is that right? So they would be a different species that doesn’t share our interests? Or, I suppose, they could have superior ethics — humans being programmed by Darwinian evolution and all. Or have I got this all wrong?
HdG: Reverse brain engineering will certainly be one of the techniques brain builders will use. Another is evolutionary engineering to create the fine level detail of the neurocircuitry. The Terrans will use, as one of their arguments against the Cosmists, that the evolved circuits would be too complex to be humanly understood and hence be unpredictable. Therefore the artilects would not be controllable by humans. Artilectual ethics would be alien to humans. Terrans will simply refuse to allow any advanced artilects to be built. They will witch-hunt out any person expressing sympathy for building artilects. God, I’m glad I’m alive now! I don’t want to see any of this horror, this Terran paranoia against the Cosmists, the cyborgs, the artilects.
H+: Ben wrote a piece for us about “the Chinese Singularity.” You’ve both been in China quite a bit, working together. Tell us a bit about how you see China moving forward in the AI field, and what are some of the values and, conversely, the problems with technological evolution in Chinese culture?
HdG: One of the advantages of being a dictatorial government, especially in a country the size of China, is that you get to do what you want, particularly when it comes to mega engineering projects, like the “Three Gorges Dam” or the new high speed rail project, for which the government simply assigned 10,000 engineers, so that China already has more high speed track than Europe.
Ben Goertzel and I are trying to persuade the Beijing government to make Xiamen (where we both live) the Chinese center for artificial brains (ABs). In the 2020s, ABs will be one of the biggest and richest industries in the world because everyone will want to spend big money on a genuinely intelligent and hence useful home robot, controlled by an AB. With Ben and me in Xiamen, we expect that further westerners will come, so that a “foreigner center” can be established that can grow and grow.
I have a rather low opinion of China’s ability to innovate. I’ve been living full-time in China now for 4 years. I don’t see China showing any real creativity (it has won zero Nobel prizes) until it becomes a democracy in about 10 years, with freedom of speech. Today’s CCP (Chinese Communist Party) is history’s greatest criminal organization, because it has killed more people than Stalin or Hitler, mostly under Mao Zedong (80 million, compared to Stalin’s 60 million, and Hitler’s 50 million). Yet his face is on the country’s money (the moral equivalent of the Europeans putting Hitler’s face on the Euro). If I were to say what I’ve just said in Chinese and to the Chinese public, I’d be hauled away to a labor camp (a “laogai”) in western China (China’s Siberia). Without freedom of speech, China’s science will remain parasitic on the west. Experience in 100 countries the past half century has taught the “transitologists” (political scientists who study the transitions from one party dictatorships to multi party democracies) that once a country reaches a standard of living of about $6000-$8000 per year per person, that is when the switch occurs. China is getting close to that level now. If the CCP is smart, it can anticipate this change and use it to its advantage, by appealing to the Chinese people — “Do you want the country run by a bunch of amateurs and see the world’s highest average economic growth rate evaporate?” Once the CCP has become the CSDP (Chinese Social Democratic Party) and has totally dissociated itself from Mao, then over a billion Chinese will feel free to say and think what they want. Then China will really become the new America. But even in a post CCP China, it will take decades for the Chinese to unlearn their deeply culturally rooted “mean spiritedness” that treats anyone outside ones social circle as a sucker to be abused. I’m predicting in the 2010s that large numbers of westerners will move to China, attracted by its growth rate, only to learn bitterly that China is still quite an unfit culture for westerners to live in. These westerners will dismiss China with the slogan “the lying Chink syndrome” the way they did in Japan in the 1990s with the slogan “the racist Jap syndrome.” I don’t see China really coming into its own by world standards until about midcentury. Then, finally, it will have become a modern, democratic, civilized culture. It still has a long, long way to go.