Creative Computations

Creative Computations
A group of engineers sits at a table discussing the design of a new cell phone. As they toss ideas back and forth, they're gesturing to show the possible width of the screen and placement of functions such as volume control. A computer watching the human gestures presents the group with visuals that approximate the potential phone's dimensions. The engineers manipulate the images—changing, saving, or discarding them as the discussion continues.

Professor of computer science David Brown envisions a time when computerized systems, through the groundbreaking field of Artificial Intelligence (AI) in Design, will assist people in harvesting their creativity in such scenarios.

Brown first discovered the world of AI in Design as a PhD student in computer and information science at Ohio State University in the early 1980s. As he considered his thesis options, his advisor suggested that he solve a design problem faced by a sensor manufacturer: the company, seeking to boost production, wanted a computer to recreate the problem-solving skills of their in-house expert.

The sensors detected thickness for machines that milled plastics and paper. For each milling application, the staff engineer crafted routine but painstaking design changes. Brown interviewed the engineer at length, essentially mining the expert's knowledge of the design process. 

Now firmly established as a pioneer in the field, he has delved ever deeper into how AI can leverage and mimic human creativity. "The field has changed dramatically over the last 25 years," says Brown, who arrived at WPI in 1985 and has held a collaborative appointment as professor of mechanical engineering since 1999. "We've gradually moved from programming routine tasks to ever more complex efforts—those involving creativity," he says. "So we've gone from solving relatively easy problems to a point where now we don't have much of a clue anymore about where we might be headed."

Brown received his appointment in mechanical engineering when he began advising ME PhD students, guiding them in programming designs for manufacturing and other fields. One of his protégés—Janet Burge '05 (PhD), whose thesis explored software engineering using design rationale—is now an assistant professor in computer science and systems analysis at Miami University in Oxford, Ohio. In 2009 Burge won the National Science Foundation's CAREER award, one of the most prestigious for young faculty members, to support her continuing research into design rationale, which explores not only the results of design processes but also the reasoning behind design choices.

While AI in Design may sound esoteric to the uninitiated, Brown sees it as a utilitarian tool, grounded in logical reasoning. Creativity, after all, "exists as a judgment relative to personal or group norms," he says. "AI in Design looks to take the subjective and make it objective and, therefore, computable. Creativity can be modeled in computer code because it includes processes involving knowledge and reasoning."

Brown's recent writings explore computational creative design. His articles began touching on the subject in the late 1970s, when he explored a natural language graphics project. More recently, Brown reported on how the AI in Design community is developing definitions of creativity with an eye toward designing and judging products. For example, Susan Besemer, founder of ideafusion, has shown that people judge the creativity of commercial products by their novelty, resolution (usefulness and user-friendliness), and style. Brown thinks that computers should also be able to do that.

Additional characteristics refine those main categories. For example, novelty refers to the use of new processes, new techniques, and new concepts in the product. It also includes the newness of the product within and outside of its field. Resolution, the degree to which the product addresses a specific need, "underscores the fact that, for products at least, new but bizarre objects aren't seen as creative." Salvador Dali's surrealist Lobster Telephone, for instance, is widely accepted as creative in the art realm. But if viewed as a product, the lobster phone would flunk the creativity exam due to its limited utility.

As such definitions emerge, Brown sees AI in Design stimulating creativity by assisting human experts while they toil at complex projects. Computational models that grasp the basic processes involved in, say, creating buildings, could provide advice and cautions about potential pitfalls in emerging designs.

Next-generation computational design systems could also stimulate creativity by alleviating tedium. Far more ambitious than Brown's PhD thesis for designing part of a sensor, he suggests that a brave new AI program might look more like an architectural grammar to help create buildings. Such a tool would be similar to a language grammar for word processing applications. But instead of piping up when sentences ramble, this lexicon of windows, doorways, materials, and other architectural elements could open imaginations to new ideas for floor plans.

Brown is excited to think about AI in Design interfacing even more actively with people. He points to a special issue of a journal he edits, Artificial Intelligence for Engineering, Design, Analysis and Manufacturing, that explores gesture. When computers are capable of interpreting human gestures within a knowledge context, such as architecture, building design discussions could evolve more rapidly or simply more creatively. "I find this area of AI in Design fascinating," he says. "Computer design aids that interact with human designers to support their activity seem to me the most practical next step in this field."

But how or when will AI succeed in computationally modeling creativity? "It's up in the air," says Brown. "AI in Design is so complex and rich that whenever I look at it, I know we're just scratching the surface."

Maintained by