The following is a special contribution to this blog by Ashok Goel, Associate Professor of Computer and Cognitive Science in the School of Interactive Computing and Director of the Digital Intelligence Laboratory at Georgia Tech. Ashok attended the AAAI Spring Symposium on AI and Sustainable Design in Palo Alto, CA, last month.
Computational sustainability is becoming big in artificial intelligence (AI) research. For the first time ever, this year AAAI is organizing a special track on computational sustainability as part of its annual National Conference on AI. In addition, this year AAAI held a spring symposium on AI and sustainable design, organized by Doug Fisher (Vanderbilt Univerity) and Mary Lou Maher (University of Maryland at College Park). The symposium was preceded by a NSF workshop on computational methods and tools for biologically inspired design, which I co-chaired with Daniel McAdams (Texas A&M University) and Robert Stone (Oregon State University). All these promising initiatives raise the question of linkage to me: What, for example, is the scientific link between AI and biologically inspired design, between AI, biologically inspired design and sustainable design, and between AI, sustainable design and computational sustainability?
Let us take these questions one at a time. Biologically inspired design, also known as biomimicry or bionics, is a growing movement in design that espouses the use of nature as an analogue for generating ideas for designing technology. Think, for example, about the design of new iridescent computer screen display technologies based on the nanostructures on the wings of morpho butterflies. While biologically inspired design is growing as a design movement, its practice remains ad hoc, with little systematization of biological knowledge from a design perspective or the design process itself. This is where AI can help — by helping answer a myriad of questions: What might be a good ontology for representing knowledge of biological systems? What might be a good ontology for representing problems of technological design? How might biological knowledge relevant to a design problem be accessed? How might design patterns be learned from specific biological analogues? What might be the process of analogical transfer of biological knowledge to design of technology? How might a design concept be evaluated? And so on. Thus, AI offers process, content and representation techniques of problem solving, memory and learning to help transform biologically inspired design from a promising paradigm into a principled methodology.
What about the relationship between AI, biologically inspired design and sustainable design? First, note that while biologically inspired design is a method, sustainable design is a problem. That is, one can potentially use the method of biologically inspired design to address a problem in sustainable design. Second, note also that not all designs generated through biologically inspired design are necessarily sustainable. The design of a robot that can walk on water mimicking the locomotion of the basilisk lizard has little to do with sustainability. However, the design of blades in windmill turbines based on the tubercles on the fins of humpback whales is clearly driven by a problem in sustainable design. But note that the transfer of the design of tubercles requires a deep understanding of their functional role in maximizing energy efficiency. Thus, if we can understand the principles of biological designs — the principles that make them energy efficient, for example — then biologically inspired design could be an important method for addressing problems of sustainable design. It follows that by helping systematize biological knowledge and biologically inspired design, AI can also help address problems in sustainable design.
Finally, let us look at the relationship between AI, sustainable design and computational sustainability. The AAAI Call for Papers for the special track on computational sustainability invites papers on sensing, modeling, analysis, prediction, control and optimization of complex systems relevant to sustainability. This is an important agenda. AI research on sustainable design, in addition, holds the promise of new kinds of complex system designs such as biologically inspired sustainable designs. AI research on sustainable design also entails designs of actions, plans, behaviors, processes and policies pertaining to sustainability. I thus call upon AAAI to explicitly include sustainable design in next year’s special track on computational sustainability.
