Most designers don’t craft algorithms or work with code on the level of an engineer or computer scientist. But designers do ask questions that make for better framing and better requirements, that unearth pitfalls and biases, and help people better interpret how algorithms affect them.
Guest post written by
Molly Wright Steenson
Molly Wright Steenson is Senior Associate Dean for Research in the College of Fine Arts at Carnegie Mellon University.»>Molly Wright Steenson is Senior Associate Dean for Research in the College of Fine Arts at Carnegie Mellon University.
What’s the role of a designer in AI? Most designers don’t craft algorithms or work with code on the level of an engineer or computer scientist. But designers do ask questions that make for better framing and better requirements, that unearth pitfalls and biases, and help people better interpret how algorithms affect them.
Here are five ways that designers make for better AI.
Designers frame problems. How do you know what to design? If you change the frame of your perspective, maybe you’d make something far better. Shelley Evenson, Managing Director of Organizational Evolution at Fjord gave this example in the 1990s: What if I commissioned you to make a vase? Chances are, you’d come up with a vessel of some sort, maybe something like a glass or a bowl that would hold flowers and greenery. But what if instead, I wanted you to come up with a beautiful way for people to incorporate plants and flowers into their lives? You might design a garden, a wildflower meadow, a new concept for a plant and flower shop. “Contemporary design has changed the questions,” she says.
Designers can expand opportunities at an early stage and flag problems.
Designers and design researchers use human-centered approaches to explore and investigate the context of a problem. When human-centered designers start a project, they begin by diving into the context around it. Say you’re designing technology and training for an automotive garage. A designer and design researcher would visit garages and understand how mechanics do their work and how managers run the garage. They’d look at the existing computers on the sales counters and in the garages, as well as the smart phones in the pockets of the mechanics’ jumpsuits. They’d look into how they get information and make decisions. By exploring the context of the problem, designers contribute to finding the opportunities to engage and the problems to solve.
By exploring the context of the problem, designers contribute to finding the opportunities to engage.
Fighting biases and making better requirements. Algorithms in the world show their biases. Take for example the soap dispenser at a Facebook office that doesn’t dispense soap to a dark-skinned hand, or the HP computer camera that tracks white faces but not dark-skinned ones. “HP computers are racist,” quipped the electronics clerk in this viral video. Or camera sensors that give an “eyes closed” alert when photographing Asian subjects. What might have happened instead if designers who understood the limitations of these sensors worked on the product requirements?
Designers can help to identify problems and biases in data sets.
Not just data visualization, but data collection. Data visualization makes it possible to understand large amounts of data by representing it visually. There are sophisticated forays into data viz that engage design, art, and computer science, as well as the efficacy of data visualization in fields ranging from medicine to journalism. But what about collecting the data? Just as designers and design researchers can investigate the context of a business problem, they can also help to identify problems and biases in data sets.

Continue reading...