The Tastemakers: Design Roles in the AI Future

March 23, 2017

AI will change how we work, but won’t replace our humanity.


Over the past several years Artificial Intelligence (AI) has crept into various corners of our daily lives and become a dominant theme in our conversations. The benefits are cautiously held in high regard: we’re getting cars that drive themselves, homes that know the conditions you like, advertisements that anticipate what triggers you, devices in hospitals that help identify and treat diseases, and more as AI develops. These are specialized skills previously thought to be off-limits to automation and AI, that are now performed just as well or better than humans. Graphic design skills are in this bucket, and there are already many tools to automate design tasks. This automation can feel alarming, and the conversation trends toward the apocalyptic or cautionary tale (the robots are taking our jobs!), but we don’t think we should worry too much about becoming obsolete.

AI vs human capacity

AI won’t replace designers entirely, anytime soon. Instead it it will change how they work, into being the people who craft the culture – creating trends, dreaming big ideas – while the machines take care of the simpler tasks and learn what they can from them.

Current applications

There are several AI design tools already alive to aid designers. For now, they automate “simple” tasks – production tasks – such as cropping and enhancing images, placing them into designs, choosing color schemes, arranging content, and assembling pre-designed elements. Netflix is using algorithms to create the featured imagery on its apps – their program can interpret an image, look for faces, bodies, or complex areas, and dynamically crop the image based on what it “thinks” is the focal point. In the same manner, it then applies localized text and graphics on the image without blocking other text or faces. Wix has a very sophisticated website creator they’re calling Advanced Design Intelligence, which goes beyond pre-designed templates and uses machine learning to “design” a unique website based on its knowledge of how other good websites look and behave. Content, images, and other elements are arranged automatically to have the best user experience, while still appearing to be a unique, individually designed website.

At first glance, this all sounds great and tangibly useful in many of our day-to-day tasks. Recently, while working on one of our client’s TV streaming apps, I found myself wishing for some such automation of my tasks. We had a detail screen with many different states and varying combinations of actions a user can take (watch live, start over, record, delete recording, etc). I found myself craving an AI tool that I could tell what to do (screen 1 will have buttons a, b, and e; screen 2 will have buttons a, c, and d, etc). I’d push the button and all my variants would be assembled and ready.

Our team is also developing a system that pulls a dominant color from an image to use for the background, with Lokesh Dhakar’s Color Thief. We found that most images processed with this tool delivered a muddy, desaturated hue. However, within the palette of colors it collected, there were others that were brighter and more saturated. It seems that there’s an algorithm waiting to happen that can scan those colors until it finds one with a diverse histogram – red, green, and blue values that are distinct from one another – and uses the color that passes the saturation/vibrancy threshold we want.

The machine & its master

This seems like the natural progression all industries take. A job is first done by a small set of skilled craftsmen who can do it properly; more and more people learn how to do it, some better than others; and at some point we create a machine that can do it for us at great speed. We’ve made machines to make chicken pot pies for us: a person programs what variety of crust to use, whether or not to add the chicken bits, which vegetables to add. What’s important is that there is still a human in there somewhere telling the machine what to do. The machine does what it knows to do, rather than having free reign to do whatever it wants and make a mess.

Likewise, the AI isn’t doing anything new. It uses skills it has learned, rearranges elements it knows how to make, composes things with the data it reads. The progressive, cutting-edge designs can only be done by humans (for now). We have the ability to process emotions and predict how a design will make a user feel. We have the brainpower to try new things and break boundaries, make new ideas work. We can watch the trends and determine what we think is ugly or well-designed.

Our intuition and ability to derive meaning from our sensate experiences allows us to interpret outcomes and behavioral responses, adding an extra layer of empathy that AI will struggle to emulate. Trevor O’Brien remarks, in his excellent article on the possibilities of artificial intelligence and designers working together:

“Humans understand concepts as they apply to the arts and creativity in a permeable way; they are able to break through the accepted rules of an art form…in a way that leads to other related concepts and gives birth to new ways of thinking. Computers, on the other hand, cannot take the information they have about a type of music and use it to transcend the form. Which is to say, if it were up to computers, we never would have evolved from the blues to rock ‘n’ roll.”

A computer’s choices will always be rational based on the hard evidence it gets its hands on.

In this mindset, we can be hopeful that design as a specialized craft has nothing to fear from AI. The simpler, production tasks will be done autonomously, and designers will be the ones teaching the computers how to “design” properly, synthesizing the things they can create, and doing the high-level tasks a computer can’t. This will obviously make our jobs more enjoyable – we can use the time we spent applying the same design to multiple surfaces, or exporting an image to seven different scale factors, to instead work on the more enjoyable things that we’re really passionate about.

Human designers will become the tastemakers, introducing next-level ideas, setting the benchmark for “good” design, teaching the machines how to work. It will necessitate us to be ever better at our jobs, so as to “compete” with the AI. How would it feel to design a website that’s inferior to one the AI designed in a fraction of the time? How will that affect your self esteem, that this job that you invested many years to hone, likely some years of university education in, is done slightly better by a robot? Though this may be somewhat inevitable in almost all industries, for now and the near future, AI stands a chance to make us better.

Human and Computer

We will have a coexistence with AI, instead of the dystopia of computers taking our jobs entirely.

The value of human hands

These automations are all well and good, so far, for designers at agencies with enough work to delegate to humans and to computers. The freelance designer may be hit the hardest once AI for design really comes into fruition. Consider a good logo designer: one can charge thousands for a logo and identity that might have taken an hour to create, for the intellect that went into it. The value is not derived by the time spent in the actual designing of the logo. Paula Scher, one of the goddesses of graphic design, tells a story of designing the Citibank logo after their merger in 1998. She sketched it on a napkin shortly after her first meeting with the executives, in one take, and that was the concept that stuck. “How can it be that you talk to someone and it’s done in a second? But it is done in a second. it’s done in a second and in 34 years, and every experience and every movie and every thing of my life that’s in my head.” The task takes a minute, but that relies on decades of experience to know how to do it in a minute. The experience and the intellect are what makes the logo great and what adds the value.

Sites like logojoy.com, tailorbrands.com, and MarkMaker already have algorithms to automate this process. I gave LogoJoy a spin to redesign the L4 Digital logo. I typed in our company name and slogan, it gave me a handful of existing logos to choose from as inspiration, I chose a color range and some symbols, and I instantly had five iterations on a logo. They weren’t fantastic, by any stretch, but passable.

 

AI Made Logos

AI-created logos are often nothing more than different combinations of a symbol + type in different treatments and colors.

 

For a client who can’t afford the full design process by an individual or a team, but can swing the $130 Logojoy charges for a full identity package, this could easily take a job away from a human (meanwhile, the creator of this service is reported to be making $70k a month from it). However, the outcome is a dreadfully basic logo, far from a piece of genius that a professional designer could dream up, synthesized from deep knowledge about the company’s values, personality, and goals.

side by side

The other side affect of this automation is a devaluing of the industry. Design is an art, and designers have a deep knowledge of what “good” design is and how to create it. Not everyone can recognize or generate these concepts. If we start using these readymade logos and websites in excess, what does that do to our craft? The same devaluation has been happening to photography since the iPhone camera came around; why would I pay a professional photographer to take photos with their high-end equipment, when I can push one button on my phone and get an image of similar quality?

Tomorrow

Artificial Intelligence won’t replace our jobs as graphic designers anytime soon. It will change us, by encouraging us to stay on the cutting edge of design trends and be among those creating the trends that computers will replicate. We’ll have more time to explore those big ideas that we’re passionate about, and spend less time on the repetitive production tasks. New roles will be created, for people who create the algorithms and machine learning practices to enable computers to replicate what designers create. We’re still in the young stages of the computing industry, though, and computers are getting smarter. A project at MIT is learning how to match images with spoken language; the creative agency M&C Saatchi made a living advertisement powered by a genetic algorithm that measures the reactions of the people looking at it, and alters its imagery, layout, and even the text content, to be the most effective piece for the person looking at it. Whether tastemaking is something you can program remains to be seen,  but in the meantime we’ll continue to place our trust in the very human capabilities in the field of design. The value lies in the fact that we’re humanly intelligent, not artificially so.

Mark Prescott

Mark is a graphic designer at L4 Digital. Before coming to L4, Mark spent several years working professionally as a designer in a range of fields including marketing, packaging, and UI/UX. His background in fine arts gives him a well rounded and unique perspective into the world of graphic design.

BLOG ARCHIVE
Before jumping on the Kotlin train, make sure your team is o...
The L4 Bug
Google I/O: AI, Right?
  • June 2017
  • May 2017
  • April 2017
  • March 2017
  • February 2017
  • January 2017
  • December 2016
  • November 2016
  • Contact L4