Exploring ARTIFICIAL Intelligence

Years ago, when associate professor of psychology Olaf Sporns was in high school, he replicated the earth's ancient atmosphere by mixing gases, including methane, hydrogen and ammonia, and esind sparks through the gases to synthesize building modes of early life. \n"One day, I blew the top off the apparatus," Sporns said. "No big deal, really. But I was more careful after that."\nSporns is still fooling around with science. He has been at IU since 2000 and is widely recognized as one of the most indispensable members of the interdisciplinary effort, or 'revolution' as he refers to it, to further knowledge and improve technology that might eventually lead to artificial intelligence. \n"I know of no other faculty, at Indiana University or anywhere else, who has so successfully conducted and disseminated interdisciplinary research that spans across physics, biology, neuroscience and cognitive science," said Rob Goldstone, director of the cognitive science program. \nThe field of AI is often associated with failure because the benefits are very long-term. In fact, many industries shy away from AI research for this exact reason. Sporns, who has been active in the field for 15 years, said research has accelerated the past 10 years because of a greater wealth of knowledge and better technology.\n"Part of my work is to take AI in a more biological direction," said Sporns. "Classical AI always had problems because they took the hardest possible approach."\nGoldstone said that since Sporn arrived at IU, he had brought both himself and the University national recognition, and his research has flourished in a number of novel and field-defining directions. \nSporns said artificial life is all about starting simple. The early attempts at AI made the erroneous assumption that they were going to be able to mimic human-level intelligence. \nSporns does research on spatial learning, memory, reward systems, motivation and attention. Goldstone said that what makes Sporns' work so worthwhile is the way he creates well-specified theories of behavior by designing working systems that implement these theories. The theories even allow robots to learn because of structures that are included in-brain. \n"It is one thing to provide a general account of perception," Goldstone said. "It is quite another thing to provide an account that is so well-specified that it can be built into a computer that is then able to perceive in human-like ways."\nSporns develops his models by building entire robotic systems, with nervous systems and bodies. Robots are actually connected to their nervous systems, which are in a computer, via a bundle of wires that allow robots to receive motor signals, connect the robot to a power source and allow researchers to record video. \nStrider, a combination of colorful wires and metal, is Sporns' favorite robot because it has the ability to move freely. Strider has four legs, each with three motors. Strider cost about $500 to build and was actually made by undergraduates, who Sporns includes in different facets of research, including co-authoring scientific papers.\nSporns demonstrates that bodies cannot be easily separated from minds. Neural organizations develop in conjunction with the bodies that contain them.\n"Realistically, it is still an uphill battle," he said. "Areas that will be investigated in the future are mental functions that are more removed from the stimulus like creativity, rational thought and language."\nSporns regularly travels to Europe and Japan because of joint projects with the University of Zurich and the University of Tokyo. \n"The Tokyo group is using our jointly developed strategies to design life-size humanoid robots with advanced sensing and movement capabilities," Sporns said. \nCollaborative efforts such as this one are becoming a distinctive part of the overall endeavor to create AI. \n"Indiana University is one of the world leaders in this revolution that's going on," Sporns said. "That's the reason I came here -- because there is a number of people who are thinking along similar lines."\nLarry Yaeger, a former Apple scientist and Informatics professor who has been at IU for close to two years, is one of those people. Yaeger said that he and Sporns both have been thinking a long time about measuring growth and development that occurs in neural systems and using those measures to characterize and drive artificial life simulations.\nSporns is responsible for creating a way to quantitatively measure how complex a given pattern of activity is in a neural network. Yaeger applied this measure to the brains of his artificial organisms that operate in a computer world, a simulative ecosystem called Polyworld. Through their collaborative work, the researchers witnessed evidence of growth in complexity in the creatures that live in Polyworld. \n"They need bigger and better brains to survive more effectively in their environment," Yaeger said. "Modeling living systems allows people to learn a great deal how complex systems come to be, how they persist and basically control the organisms." \nNew researchers are being recruited to strengthen an already internationally recognized faculty and to build undergraduate education. Randy Beer will start work this fall in Informatics and cognitive science. \n"Within the field of cognitive science, there is a growing interest in understanding, not only purely intellectual capabilities, but also how a cognitive system exploits its body and its environment in its cognition," Beer said. "The cognitive science program at IU has a long history of interest in such situated, embodied and dynamical approaches to cognition"