Challenging Students to Think: Smarr Presents Radical Scenarios of the Future
February 21, 2006 -- Calit2 director Larry Smarr was invited to offer some of his views on future trends in a lecture titled "The Singularity: Toward a Post-Human Reality" to UCSD Sixth College's Sophomore Honors Seminar last week. The seminar series, whose topic is different each quarter, focuses on analyzing Ray Kurzweil's recent book The Singularity Is Near . Smarr's lecture was one by a series by futurists, including Verner Vinge, who coined the term "singularity." (Vinge is a professor in the Department of Mathematical Sciences, San Diego State University, and an acclaimed science fiction writer.)
|
The provocative thesis of Kurzweil’s book is that “because technological change is exponential, we won’t experience 100 years of progress in the 21st century -- it will be more like 20,000 years of progress (at today’s rate). Within a few decades, machine intelligence will surpass human intelligence, leading to The Singularity -- technological change so rapid and profound it represents a rupture in the fabric of human history. The implications include the merger of biological and nonbiological intelligence, immortal software-based humans, and ultra-high levels of intelligence that expand outward in the universe at the speed of light.” (See, for instance, www.kurzweilai.net/articles/art0134.html.)
Smarr began by pointing out that the use of the term “singularity,” as used by Vinge and Kurzweil, is metaphorical, not physically true, since a physical singularity is a point in the future when a value appears to go to infinity. If one plots an exponential function on a linear scale, it will appear that the value of the dependent variable gets very large as time increases. However, that is simply the exponential crashing through a pre-established linear threshold. As Smarr quipped, “an exponential turns the ‘impossible’ into the routine.”
But Smarr advised that Kurzweil’s approach of keeping one’s eye on a variety of “exponentials” is one of the best predictors of the future. He used the example of Moore ’s Law, a classic principle in the computing world that says that computers have doubled and will continue to double their speed every 18 months, such that a 1990 “supercomputer” costing $15 million is exceeded in capability by the inexpensive PCs that many people carry around as laptops.
Exponentials, though, only become visible to people when they cross a widely perceived threshold. Again, drawing from recent history, Smarr cited the telephone system, which had experienced 3-4% growth in traffic every year predictably for decades. What the phone companies didn’t bargain on, though, was the emergence of the Internet whose traffic grew at a rate of 100% per year. By the year 2000, the Internet curve crossed the telephony curve (when Internet traffic equaled phone voice traffic). By the end of the decade, 95% of the traffic will be Internet, with only 5% voice.
“Telephone companies needed to spend billions to build systems that could carry all that extra Internet data traffic – but with no additional revenue,” said Smarr. “It’s no wonder companies started failing: They just couldn’t read the exponential curve.”
Technology diffuses into society following an S curve, said Smarr. The classic S curve can be described as consisting of three parts: innovation, growth, and maturity. Innovative research and development can last a long time (the flat lower portion of the S curve) while pioneers “play” with the new technology to figure out how to apply it. Gradually, the results of these experiments are engineered into products that “early adopters” begin to buy, thereby starting the move up the S curve. After 5-10% of the population has the product, the “buzz” growth phase begins when the broad population “discovers” the new product, and uptake becomes very rapid. When something like 90% of the people have the “new” technology, the product reaches maturity, the rate of growth slows toward zero, and the curve flattens out (top portion of the S).
Calit2 focuses on the first phase of innovation when most people still don’t “get it,” and the technology is less robust.
“Calit2 is designed to study the types of trends you read about in Kurzweil’s books in a cross-disciplinary way,” he said. “Calit2 faculty and staff write grant proposals that fund us to deploy early examples of technology and integrate them into what we call ‘Living Labs.’ Why would we want to do this? It’s simple: If you’re working on an exponential and can see 3-5 years into the future, you get an ‘early-warning system’ of potential new applications. Our new facilities at UCSD and UCI give us the opportunity to do this kind of experimentation – ‘living in the future.’”
Smarr encouraged the students to read science fiction as it provides views into possible futures. “One of the purposes of art is to imagine how the world might be different in the future,” said Smarr.
Smarr pointed to several themes that science fiction has embraced, all of which are relevant to the issues Kurzweil discusses:
- “Software-augmented or Transhumans,” which play central roles in Aristoi by Walter Jon Williams. In this book, artificial intelligence agents communicate with their human host, giving the person a feeling of a split personality. “You have multiple special agents living on the net. They have personalities and fight with each other -- just like real people,” said Smarr.
- “Posthumans,” in which the corporeal body is left behind, and the pattern of consciousness on the individual’s “wetware” is injected into “cyberspace.” There are classic books (e.g., Neuromancer, Count Zero, and Mona Lisa Overdrive, all by William Gibson), TV shows (“Kill Switch” in X-Files’ fifth season), and movies (the recent Matrix trilogy) that describe such Post-Singularity possible Earths.
- Intelligent robots and androids. These have been central characters in science fiction, from the book series on robots by Isaac Asimov to movies such as Blade Runner and the Terminator trilogy.
Smarr then turned back to Kurzweil to discuss several of his “accelerators” in the context of Calit2 research and development activities.
“The first accelerator is what I call the ‘perfect storm’: the convergence of info-, bio- and nanotechnology,” said Smarr. He illustrated this convergence by showing images of two nano-scale objects: IBM’s “quantum corral” and Nature’s rhinovirus. Smarr observed that though one might consider the former “physical nano” and the second “biological nano,” both were just structured sets of atoms that contain information, a “1” or “0” inside the corral (a nano-memory) and RNA inside the virus (a nano-program). This ability of a nano-scale object to carry out computing functions is central to many of Kurzweil’s conclusions, Smarr observed.
Calit2 has built clean rooms in its buildings at UCSD and UCI to support long-term research in nano-science, nano-engineering, and nano-medicine. For instance, UCI has assembled a world-class program in micro- and nano-biology. UCSD has a large project to study the creation of nanosensors that can measure physical, chemical, and biological properties of the human body. In addition, an interdisciplinary team led by UCSD’s professor Sadik Esener recently was awarded one of five NIH-funded grants in applying nanotechnology to the treatment, understanding, and monitoring of cancer.
Human longevity was the next topic. Kurzweil observes that, with all the new advances in medicine, some undergraduates today may have radically longer lives than their parents. Indeed Kurzweil has written another book, Fantastic Voyage, in which he lays out diet, exercise, and meditation modifications to the human life style that could help people “live long enough to live forever.”
Calit2 has a variety of affiliated research programs studying the basic properties of living creatures, which could support this medical revolution. UCI professor Douglas Wallace has a center for mitochondrial diseases. (Mitochondria are the power sources inside each cell.) Calit2’s Eleazar Eskin has recently collaborated with academic and industrial partners to study variations in the genome across the human population. In both cases, genetic subclasses are often linked to susceptibility to certain diseases.
“We’re starting to see the development of genetically tuned drugs that are tailored to the genetic constitution of the individual,” said Smarr. “But for this to become commonplace, pharmaceutical companies will need to give a lot more thought to new business models.”
The next accelerator is Kurzweil’s notion of reverse engineering of the brain. This led Smarr to talk about his large National Science Foundation-funded OptIPuter project (www.optiputer.net), which has helped support development of 100-megapixel tiled display walls by the Electronic Visualization Laboratory at the University of Illinois at Chicago . These displays can show the global context of a data set and allow the user to “drill down” in increasingly fine resolution on areas of greatest interest.
He illustrated this using a 300-million pixel image of a slice of rat brain produced at professor Mark Ellisman’s National Center for Microscopy and Imaging Research, a Calit2 partner funded by the National Institutes of Health in UCSD’s Medical School . The image was created by a robotically controlled confocal light microscope that can collage microscope images into a seamless large-scale mosaic covering the entire brain slice. (By contrast, a typical PC screen can show just 1-2 million pixels.)
“Our OptIPuter technology allows zooming from the cerebellum down to individual neurons,” said Smarr. “This sort of computer enhanced imaging technology, which gets better as computers get faster, is one source of optimism for Kurzweil’s belief that scientists will be able to gradually understand the brain’s wiring over the next few decades.”
However, Smarr cautioned that even if we understood the macro neuron connection map, we still need to model the individual cells and the complex information flows that occur within and between these cells. This is the justification behind the “visible cell” project, led by Ellisman and housed in the Calit2 building at UCSD. A cell-centered database will collect data on cellular sub-components and cellular processes in three dimensions and over time. The first generation of distributed computing simulations for cells then uses these data to predict cellular behavior. However, Smarr warned the students not to underestimate the complexity of living cells. He pointed out that a single cell not only has 3 billion bases along its nuclear DNA, but it has 4 million ribosomes that carry out protein synthesis, producing some 5 billion proteins drawn from 5-10 thousand types. So reverse engineering the brain is a daunting, multi-decade challenge.
Then it was on to how to make computers faster than human brains. Smarr noted that while brains are very powerful, they must have a finite limit to their speed. As Kurzweil, Hans Moravec (author of Mind Children: The Future of Robot and Human Intelligence) and others have estimated, this speed limit is roughly one million billion operations a second, also known as a Peta-op. While this sounds incredibly fast, Smarr noted that modem PCs can do several billion operations per second, and there are close to a billion of them in the world today. “So if we just hooked up ten million PCs via the Internet, that collective raw computing power would theoretically exceed that of a human brain,” he said. For the last five years, numerous projects have been hooking up PCs to create planetary-scale, distributed computing capability for a given problem, the most well known being the SETI@Home project centered at UC Berkeley.
As the planetary computer is passing through this important threshold of the speed of a human brain, will it become self-organizing, -powered, and -aware? (See Vinge’s speculations on this at www-rohan.sdsu.edu/faculty/vinge/misc/singularity.html.) “If you have enough computing power, maybe computers will start writing their own software,” said Smarr. That will mean that “software as engineering” as we know it today might become “software as biology,” again pointing to the convergence of info-, bio-, and nanotechnology.
With that kind of planetary computing “power grid,” robots could use this huge computing power to augment their “on-body” computer brain. Already commercial robots from Sony have WiFi built in, so such distributed processing is possible in principle. Such off-body computing power, millions of times more powerful than the internal computing that robots have used traditionally, could vastly accelerate the speed with which they become self-aware. Smarr noted that Calit2 researchers in the UCSD MPLab have a robot called RUBI that interacts with pre-schoolers, detecting six basic facial expression and distinguishing voices by combining spatial and temporal processing just using on-body computing. What could it do with a million times this much computing power?
Finally, Smarr turned to the topic of using far more of the planet’s human potential as an accelerator. He showed maps that clearly indicate how few of the Earth’s people produce innovations in science. “What if we could use all of the human brain power on Earth to discover -- that would be a huge accelerator,” Smarr said.
This led to some thoughts on the impact of global communication. For instance, a new project involving Calit2, the UCSD Jacobs School of Engineering faculty, and other universities and companies in the U.S. will teach Indian students remotely. “This will ‘short circuit’ India ’s need to spend generations to produce enough scholars to train large numbers of Indian students. Using global communications systems, professors in the U.S. can expand their classes to include remote Indian universities,” said Smarr. Another example in which undergraduates play the lead role is the NSF-funded PRIME program in which UCSD undergrads study abroad with professors in Australia , Japan , Taiwan , China , and Thailand . “The students use the Internet to ‘glue together’ the U.S. and foreign researchers, and come out of the program trained as ‘global scientists,’” said Smarr.
Smarr also talked about some of the technologies being prototyped by Calit2 to link human beings together “as if distance had been eliminated.” Specifically, he described an experiment at the global iGrid2005 workshop last fall using a super-high-definition projection capability that is the only one of its kind in the U.S. “This technology goes beyond what we think of as videoteleconferencing to the beginning of true telepresence,” he said.
Alluding to Thomas L. Friedman’s book The World Is Flat, Smarr said, “Today’s innovations in terms of dedicated fiber paths, streaming high-definition TV, ubiquitous wireless Internet access, location-aware software, and sensornets will take the ‘flat world’ we’re beginning to appreciate today and reduce it to a single point in 10 years. We are going to effectively eliminate distance.”
This seminar series, for high-achievers, challenges the students with the “big issues” of today.
Said
(Powerpoint slides of Smarr’s presentation are available at http://www.calit2.net/newsroom/presentations/lsmarr/index.php.)