Keynote Speakers
Confirmed Speakers
Cynthia Breazeal, MIT Media Lab
Mechanisms of Social Learning applied to Human-Robot Interaction
Abstract: New consumer applications for robots are motivating research into building autonomous robots that play a beneficial role in the daily lives of ordinary people. This "new breed" of personal robots must be natural and intuitive for the average consumer to interact with, communicate with, work with as partners, and teach new capabilities. In particular, the ability for robots to learn new skills, goals and tasks from interpersonal interaction poses new challenges and opportunities for machine learning systems. How do people want to teach robots, and what implications does this pose for robots that learn from people? Results from our human subject studies indicate that this process is profoundly social and collaborative in nature. Furthermore, social-cognitive skills (such as shared attention and perspective taking) coupled with tight multi-modal communication play an important role in the teaching/learning process. In this talk, I shall present these empirical findings, describe our computational models, and illustrate their application to robots that learn from people via social referencing, imitation, demonstration, and tutelage.
Bio:Cynthia Breazeal is internationally known for seamlessly blending scientific theories, artistic insights, and engineering principles to create compelling robotic creatures that have a lively and compelling social presence to those who interact with them. Her current research extends these themes in the area of human-robot relations to create cooperative and capable robots that can work and learn in partnership with people. Her research program strives to revolutionize the art and science of human-robot interaction and cooperation --- to develop robots that engage us as helpful partners that will ultimately play an valuable, rewarding, and unprecedented role in the everyday lives or ordinary people. Cynthia Breazeal is the author of Designing Sociable Robots and Biologically Inspired Intellilgent Robots
Douglas Hofstadter, Indiana University
The So-called Singularity: An Onrushing Tsunami, or Another Y2K?
Abstract:In the past few years, a number of futurologists, extrapolating on the basis of many interrelated exponential curves such as Moore's Law, have come to the conclusion that computer intelligence is rising so swiftly that quite soon, it will inevitably reach and then surpass human intelligence, and that at that monumental juncture in the history of this planet, humanity will be eclipsed and replaced by its own creations. Within a few decades, these cyberprophets proclaim, we humans will be living among superintelligent entities that are just as incomprehensible to us as we are incomprehensible to bacteria, and the upward spiral will continue from there on without limit, resulting in entities "who" are literally billions of times more intelligent than today's humans are, and "who" will soon commandeer stars and then whole galaxies, finally turning the entire universe into one single inconceivably intelligent self-reflective organism akin to the Omega Point of the mystic Jesuit philosopher Teilhard de Chardin.
More modestly, these same futurologists also foresee, within only two or three decades from now, the total reverse-engineering of the human brain, as well as the complete mapping of the brains of specific individuals, followed by the uploading of those abstract patterns into cyberspace, where they will run as software and the former owners of those brains will thus become literally immortal. The futurologists also foresee nearly all biological humans choosing to abandon their biological incarnations in order to become immortal and to do eternal digital dances in virtual universes in the gigantic memory banks of hypercomputers.
Such mind-boggling visions of the very near future, eloquently portrayed in a number of well-received books by such world-renowned and internationally respected authorities as computer scientist Ray Kurzweil and roboticist Hans Moravec, sound outlandish and repulsive to many laypeople, as well as to many scientists and philosophers, while to others they seem not only plausible but also hugely exciting and full of promise. Whether it is crazy or not, this "singularity scenario" has been given spirited scientific-sounding defenses in a growing number of books by serious thinkers.
Is this "Rapture of the Nerds", as it has been called, really going to take place within our lifetimes, or are such ideas just pie in the sky? Are we all going to be swept away by an unimaginably powerful tsunami of computational brute force, or is the whole vision just going to fizzle pathetically when the predicted crucial dates arrive and nothing at all happens -- "Y2K II", as C3PO might call it? Or can anyone say?
Is it in fact possible to gain any kind of objective sense of the realism or nonrealism of these possibly loony, possibly sane, perhaps ecstatic, perhaps apocalyptic claims? The problem is that many disciplines are involved, ranging from physics to evolutionary biology, neurology, artificial intelligence, psychology, computer science, nanotechnology, molecular biology, philosophy of mind, ethics, and more. No one person has a sufficient mastery of all these fields to be able to see clearly into the future -- not even a couple of decades, let alone a century or a millennium. This means that no one today knows the truth of such claims.
In my talk, despite having no crystal ball, I will give a personal reaction to, as well as an attempt at an appraisal of, these mystic, hubristic, surrealistic, futuristic, idealistic, optimistic, spiritualistic, pessimistic, materialistic, sensationalistic, computeristic scenarios in which biological and artificial life swirl ever more intimately together in ways that go far beyond the wildest dreams of all but the most fanciful of science-fiction writers and the most zany of scientists.
Bio:For the past three decades, Douglas Hofstadter has directed the Fluid Analogies Research Group (FARG), whose collective goal has been the design and implementation of computational models of creative analogical thinking and of its subcognitive substrate -- namely, fluid concepts. FARG-style architectures -- neither symbolic nor connectionist -- are characterized by parallelism, randomness, and emergent behavior. Hofstadter is also the author of several books, including Gödel, Escher, Bach (Pulitzer Prize, 1980); The Mind's I (co-edited with philosopher Daniel Dennett); Metamagical Themas (a collection of his Scientific American columns); Fluid Concepts and Creative Analogies (co-authored with FARG members); Le Ton beau de Marot (an exploration of translation, creativity, language, and beauty); and a verse translation of Alexander Pushkin's sparkling novel-in-verse Eugene Onegin, considered by Russians to be the fountainhead of their literature. Hofstadter has recently translated Françoise Sagan's novel La Chamade into English, and is now putting the finishing touches on two new books: I Am a Strange Loop (on the elusive concept of "I" and the mystery of consciousness) and Capitals in Capitals (a collection of some of his recent ambigrams).
Clyde A. Hutchison III, Venter Institute
The prospect of synthetic life
Abstract: Approaches to the synthesis of biological life in the laboratory will be discussed. These range from the bottom-up design of primitive "RNA life" forms, to the duplication of modern cells by synthesis.
The synthetic biologist views the DNA genome of a modern cell as its operating system. The ribosomes and other parts of the translational and transcriptional apparatus are the hardware that caries out the instructions contained within this OS. Traditional genetics has allowed us to alter these instructions in useful and informative ways, but has limited us to manipulation of existing sequences. With a synthetic genome this limitation disappears. Dramatic progress is being made in our ability to synthesize large DNA molecules chemically. Designing and building synthetic genomes that function properly will be a true test of our understanding of the cell at the molecular level. We expect that dramatic alterations of gene content will become possible, and that totally novel designer genes can be included.
At the Venter Institute we are designing and building a synthetic version of the Mycoplasma genitalium genome. With only some 500 genes, this is the simplest known cell capable of independent growth and replication. We divided the chromosome into gene cassettes, each of which is being built from chemically synthesized oligonucleotides. We plan to test genomes assembled from such cassettes for functionality by "genome transplantation", replacement of a cell's resident chromosome by a new one.
Genomics has given us an unprecedented view of life's underlying code. Synthetic biology promises us the power to rewrite it.
Bio:Clyde Hutchison has carried out investigations on biological systems ranging from bacteriophage to mice. The unifying theme has been a continuing search for improved methods to learn about gene function from DNA sequence information. He has been involved in genomics since before the advent of modern DNA sequencing.With Marshall Edgell, Clyde Hutchison dissected the genome of phage phiX174 with restriction enzymes in the 1970's, and was a member of the team in Fred Sanger's lab that sequenced the phiX174 genome; the first DNA molecule completely sequenced. Since that time he has been interested in a variety of problems in viral, bacterial, and mammalian genomics. Among his current projects is a collaboration with Ham Smith, Cynthia Pfannkoch, and Craig Venter on synthetic genomics. They have been improving methods for the assembly of large DNA from chemically synthesized oligonucleotides, and proven these methods by assembling a synthetic bacteriophage phiX174 genome.
Hod Lipson, Cornell University
From Virtual to Physical Artificial Life
Abstract: One of the stated goals of Artificial Life research is the re-creation of biological life in synthetic media - machines that can adapt their behavior and morphology over evolutionary and lifetime scales. While our understanding of these adaptation processes in abstract simulations is becoming increasingly detailed, our ability to realize these processes in physical media is still very limited. Many challenges remain - from 'crossing the reality gap' between simulation and reality, to finding non-biological physical media that can accommodate morphological development, variation and replication. This talk will outline both algorithmic and hardware approaches to addressing these challenges as well as currently open problems.
Bio: Hod Lipson is an Assistant Professor at the departments of Mechanical & Aerospace Engineering and the faculty of Computing & Information Science of Cornell University in Ithaca, NY. Prior to this appointment, he was a postdoctoral researcher at Brandeis University's Computer Science Department and a Lecturer at MIT's Mechanical Engineering Department. Lipson's research focuses on new methods for autonomous adaptation in behavior and morphology of robotic systems, with broader impacts to design automation and manufacturing technologies. His work uses primarily biologically-inspired approaches, as they bring new ideas to engineering and new engineering insights into biology.
Melanie Mitchell, Portland State University
Four Principles of Information Processing in Complex Adaptive Systems
Abstract: It is still largely a mystery how complex adaptive systems in nature perform sophisticated information processing. Such systems consist of large numbers of simple, locally interacting agents with no central control. The agents collectively process information so as to recognize external and internal patterns, monitor the system's overall state, and produce behavior that is adaptive for the entire system.
In this talk I will survey what is known about information processing in several natural systems, including the immune system, insect colonies, and cellular metabolism, and I will abstract four principles of information processing common to all these systems. These principles concern the encoding of global information, the interplay of "top-down" and "bottom-up" processes, the need for a fine-grained parallel architecture, and the essential role of randomness.
I will also discuss how these principles can guide us in building artificial systems with adaptive, life-like, and intelligent behavior.
Bio: Melanie Mitchell received a a Ph.D. in Computer Science from the University of Michigan in 1990. She has held faculty or research positions at the University of Michigan, the Santa Fe Institute, Los Alamos National Laboratory, and the OGI School of Science and Engineering at the Oregon Health & Science University. She is currently Professor of Computer Science at Portland State University. Her research is on artificial intelligence, machine learning and evolutionary computation, cognitive science, and complex systems. She is the author or editor of four books and over sixty research papers in these fields.
Norman Packard, ProtoLife
The Role of Evolution in Artificial Life
Abstract: Evolution will be used as a focal point to discuss the endeavor of Artificial Life. A few examples will be given to explicitly highlight its role in determining the boundary between living and non-living systems, both in simulation and in the laboratory. We will argue that evolution and evolvability are key determining aspects for successful artificial life, and discuss the ontological status of artificial life in this light.
Bio:Norman Packard has been involved with Artificial Life since its onset and has also worked in the areas of chaos, learning algorithms, predictive modeling of complex time series, and complex adaptive systems. Norman holds a B.A. from Reed College (1976) and Ph.D. in Physics from University of California at Santa Cruz (1983). After post-docs at IHES (Bures-sur-Yvette) and IAS (Princeton), he joined the physics department at the University of Illinois, Urbana-Champaign in 1987, where he became an associate professor before leaving to become a co-founder of Prediction Company in 1991. He is currently working in a new scientific and business direction based on development of evolutionary chemistry in programmable microfluidic technology, and is co-founder of a newcompany, ProtoLife, which aims to develop these ideas in the private sector. Packard has had a long-standing involvement with the Santa Fe Institute on its external faculty. For the past seven years, he has served on the Science Steering Committee of SFI, most recently as its chairman.
Rudolph Raff, Indiana University, Bloomington
Evolution and Development
Bio: Rudolph Raff is Distinguished Professor at Indiana University's Department of Biology and Director of the Indiana Molecular Biology Institute.