The website for the Department of Defense Advanced Research Projects Agency (DARPA) does put it a little more scientifically than that (the project is officially called “Biologically-Inspired Cognitive Architectures”), but the bottom line is that they’re designing systems to that let machines “think” as humans do.
Teams from MIT, Carnegie Mellon, Harvard, the University of Maryland, and a number of other schools are developing systems intended to simulate various regions and functions of the brain. Some of these systems seem to be intended to help researchers understand the workings of the human mind by mirroring it, while others are geared to helping machines learn from experience. Many projects focus on memory and learning, as well as understanding images and words. One cleverly-named project, Learning Generalized Image Schemas for Transfer (L-GIST) aims to help machines remember the general “gist” of an event, as humans do, rather than capturing every detail. Other projects have titles like “New Minds Initiative,” “CHIP: Comprehensive Human Intelligence Project,” and “An Integrated Architecture for Grounded Intelligence in its Developmental, Experimental, Environmental, and Social Contexts.”
The DoD seems to have some serious plans for this research. From the DARPA website: “By modeling the flexibility and richness of human cognition, BICA will provide the first steps towards highly capable autonomous systems that could help the warfighter in support functions that are both manpower-intensive and dangerous.” One MIT project aims to “create a model capable of multiple levels of cognition and reflection that range from instinctive behavior to socially-conscious thinking.” Certainly, it may be a while before we see concrete results from this theoretical research, especially considering that the last call for research proposals was in January of this year. But still – robot brains! For fighting machines! Can SkyNet or the Terminator be far behind?
– Ryan DeBeasi, rdebeasi@networkworld.com
