Designers find better solutions with IT assistance, but sacrifice the creative touch

From creating software to designing cars, engineers face complex design situations every day. “Optimizing a technical system, whether to make it more usable or more energy efficient, is a very difficult problem! explains Antti Oulasvirta, professor of electrical engineering at Aalto University and the Finnish Center for Artificial Intelligence. Designers often rely on a mixture of intuition, experience, and trial and error to guide them. Besides being inefficient, this process can lead to “design fixation,” focusing on familiar solutions while new paths remain unexplored. A ‘hands-on’ approach will also not scale to larger design issues and is highly dependent on individual skills.

Oulasvirta and his colleagues tested an alternative computer-aided method that uses an algorithm to search a design space for the set of possible solutions given multidimensional inputs and constraints for a particular design problem. They hypothesized that a guided approach could produce better designs by analyzing a wider range of solutions and balancing human inexperience and design fixation.

With collaborators from the University of Cambridge, the researchers set out to compare traditional and aided approaches to design, using virtual reality as a laboratory. They used Bayesian optimization, a machine learning technique that both explores the design space and points to promising solutions. “We put a Bayesian optimizer in the loop with a human, who would try a combination of parameters. The optimizer then suggests other values, and they proceed in a feedback loop. It’s great for designing virtual reality interaction techniques,” says Oulasvirta. “What we didn’t know until now is how the user experiences this type of optimization-focused design approach.”

To find out, the Oulasvirta team asked 40 novice designers to participate in their virtual reality experience. Subjects had to find the best settings to map the location of their real hand holding a vibrating controller to the virtual hand seen in the headset. Half of these designers were free to follow their own instincts in the process, and the other half received designs selected by the optimizer to evaluate. Both groups had to choose three final designs that would best capture the accuracy and speed of the 3D virtual reality interaction task. Finally, subjects reported how confident and satisfied they were with the experience and how much in control they felt of the process and final designs.

The results were clear: “Objectively, the optimizer helped designers find better solutions, but designers didn’t like to be hand-held and controlled. It destroyed their creativity and sense of agency,” Oulasvirta reports. The optimizer-driven process allowed designers to explore the design space more than the manual approach, leading to more diverse design solutions. Designers who worked with the optimizer also reported less mental demand and effort in the experience. In contrast, this group also scored lower on expressiveness, agency, and ownership, compared to designers who experimented without a computer assistant.

“There is definitely a compromise,” says Oulasvirta. “With the optimizer, designers came up with better designs and covered a wider set of solutions with less effort. On the other hand, their creativity and their sense of ownership of the results were reduced. These results are instructive for the development of AI that assists humans in decision-making. Oulasvirta suggests that people should be involved in tasks like aided design in order to retain a sense of control, not get bored, and better understand how a Bayesian optimizer or other AI actually works. “We’ve found that inexperienced designers, in particular, can benefit from an AI boost when engaging with our design experience,” says Oulasvirta. “Our goal is for optimization to become truly interactive without compromising human action.”

This article was selected for an honorable mention at the ACM CHI Conference on Human Factors in Computing Systems in May 2022.

Source of the story:

Materials provided by Aalto University. Note: Content may be edited for style and length.

Comments are closed.