Machine Learning: How Tech Is Changing Design

image

Merel Bekking’s ‘brain-scan’ chair

Computer-aided design, or CAD, programs are as new to the world as bell-bottom pants and disco. Architects and designers started trading in their mechanical pencils and drafting tables in the 1970s – around the same time computerized dating started to vie for the place traditionally held by boozy nightclubs and well-meaning matchmakers (hi, grandma).

These days, though, the technology has been updated so drastically that it would be hard to compare the current incarnations to its predecessors (it would be a bit like putting a Tesla next to a Pinto). More than merely assisting creative professionals draw out their ideas, software programs are now helping generate the very ideas and products themselves. Computers are coming up with building layouts, package designs and furniture that are as creative or better than what humans can envision on their own.

“It’s a radical departure from what we’ve been using for the last 40 years,” says Francesco Iorio, director of computational science research at Autodesk, which develops CAD software. Later this year, a program that Iorio has been working on called Generative Design will hit the market, and, according to him, will act more like “an actual partner” in the design process rather than a passive tool. In effect, designers will be able to ask the software questions and get optimal answers back.

The program has already produced a muscular, Gaudi-esque chair called the Elbo.

Rather than coming up with the shape of the seat themselves, a design team used the software to determine the best structure given certain parameters – height, material, loads.

The legs and arms mimic forms found in nature, such as bones, which have been optimized through evolution to withstand the forces of the world. In essence, the program came up with a design “that was most fit to survive,” says Iorio, by learning from the world around it.

“The results can be surprising,” says Iorio, “because the program isn’t constrained by biases.”

Such algorithm-based software is also a way of developing mass-customized goods – broadly available items that are uniquely different for each shopper.

For example, Nutella, in partnership with HP, recently used an algorithm to generate more than seven million unique package designs to be sold across Italy.

Each one is singular, though they share a similarly jubilant aesthetic – a bit like someone has taken close-up photos of confetti as it has fallen through the sky.

It would have taken a massive team of designers an impossible amount of time and mental energy to achieve the feat.

But “the program has no limit,” according to Lavinia Francia, client creative director at Nutella’s ad agency, Ogilvy & Mather, Italy, which oversaw the project.

“It starts choosing one out of four different textures and it zooms in on it or out and/or rotates. Then it crops the selection and creates a unique sleeve. So the number of unique labels is technically infinite.”

That said, there’s still a place for people in the process.

To ensure that no meme-worthy, phallic shapes unintentionally made it onto a child’s sandwich spread, “a Nutella employee checked on every jar,” Francia says.

And a “check was made on every pattern that was mixed by the algorithm to make sure the final result would be appropriate.” (The program was so popular that all seven million jars sold out within a month).

Architect Alexis Rivas also believes there is an important, enduring role for people to play in algorithm-generated designs. He’s the co-founder of Cover, a Los Angeles-based company that builds custom backyard studios, cottages and pool houses using algorithms and robots.

“Many people’s first instinct is the fear of computers taking over all of our jobs,” he says. “But the software we use helps our team put all our time and effort into well-considered details, and the touch and feel of our spaces.”

Rivas, along with his lead designer Thomas Heyer, have devised a way, using a proprietary software, to take the desires of their clients (captured in a questionnaire) and generate a fully articulated plan in as little as three days. “We have worked closely with the guys optimizing the software,” Heyer says, “to design a set of building blocks – fixed details, how corners come together and integrated storage. Those details are taken by the software as Lego blocks and assembled into a custom design.”

One of the benefits of this kind of technology-enabled standardization is that it brings the price of the design down. Cover’s initial consultations cost less than the price of an iPhone and the structures start in the low six figures, despite the sharp, California aesthetic more commonly associated with million-dollar homes in the Hollywood Hills.

“That’s the beauty of the tech available to us today,” Heyer says. “It makes high-quality design accessible to a lot more people.”

Dutch designer Merel Bekking isn’t just interested in using technology to make high-quality design, but design that is “technically perfect.” And instead of algorithms, she uses machines that help her get directly into the minds of those she is designing for – literally. She uses MRI scanners to access the desires that are trapped deep within our brains.

“The reason I use MRIs is because I wanted to know what people really think,” Bekking says. “I know that if you ask people questions they are always prone to give socially desirable answers or maybe they don’t really know what they like, and so on. But if you put people in MRI scanners, you look at how their brain reacts,” without a filter.

For a recent project, she used MRIs of one of the world’s top design editors – Marcus Fairs, who founded the popular website Dezeen – to create a chair that was perfect for him. “Marcus was shown pictures of different materials, shapes, objects and colours,” Bekking says, “while his brain activation was measured using a 3 Tesla MRI-scanner.”

From the experiment, Bekking learned that Fairs’ brain “had a preference for orange, for closed, round shapes, for plastic and for chairs,” Bekking says. “But these ingredients were all loose ingredients. They still had to be put together.” So Bekking put together what looks like a giant, orange pill pierced on a stick, cracked open so Fairs could perch in the middle.

Curiously, though, Fairs did not like the chair, asking Bekking to take it away from his house shortly after she delivered it to his London home. “The research results were completely solid,” she said, but “as soon as he realized he had to defend to others that this is what his subconscious likes, he really started to hate it. I think this is really fascinating.”

For Bekking, using technology to create a scientifically perfect design process has also left her with a curious reaction: “Forget all the target groups, forget numbers, scientific research and big data,” she says. “I think you should trust your designer’s instinct and make beautiful things because you really feel your ideas, not because you think it will please most people.”

This piece originally appeared in the Globe and Mail on Thursday, July 20, 2017.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s