AI-Powered Design Tools
Creativity, Efficiency, and the Democratization of Design
By Dirk Knemeyer and Jonathan Follett
What impact will AI have on the practice of design? We explore this question with Tatiana Mejia, Head of AI Product Marketing and Strategy at Adobe, and one of Silicon Valley Business Journal’s 2018 top Silicon Valley women in AI. Meija takes us into the world of Adobe Sensei, the company’s artificial intelligence and machine learning technology, that is impacting all of their key software applications.
Adobe has been the leading provider of software tools for designers for over 30 years. For the designers who rely on Adobe’s software, the core materials of their work range from paper, pixels, and code, to fonts, color, and layout. In the decades before software became part of the toolset, these people were called graphic designers. Analog processes were manual, slow, and laborious. Graphic designers made progress at the speed of these physical limitations, not the speed of their minds.
The process of design has changed drastically as digital processes replaced analog and physical methods. “I remember in college going into the dark room to process my photos,” says Mejia. “And I had to be very thoughtful about what was my intent. I wanted it to look dreamy. I wanted some parts more blown out than others. Often, it took multiple approaches to it. So, if you think about that activity of taking the photograph and turning it into art or an expression, now we do that mostly digitally. And in a way, that’s automation.” As design shifted over to the digital, with tools such as Photoshop and Illustrator, things began to speed up. Making a comp shifted from an arts-and-crafts exercise to the mouse, monitor, and keyboard. Subsequent innovations and improvements have made this faster, and faster, and faster.
[Photo: by Markus Spiske on Unsplash]
As software and the Internet evolved not just the tools but also the media within which to apply design, a complicated array of titles and digitally-native identifiers emerged. These are now often generically encapsulated in a broader discipline, that includes researchers, engineers, and other specialties beyond designers, called user experience. Since the rise of mobile computing a little over a decade ago this particular design profession has exploded. This is reflected in the growth of Adobe itself. In 2005, a year before the release of the iPhone, Adobe had under 3 billion dollars of annual revenue. By 2018 that had risen to more than 9 billion dollars. While this is the product of various factors, it is also a reflection of how many more people are using these digital design tools in the world.
Given Adobe’s size, influence, and entrenched position as the market leader for digital design software, it should be no surprise that they are also investing heavily in machine learning and artificial intelligence as tools to improve their offerings. And while, on one hand, the established nature of their software provides some constraints as to what sort of solutions can realistically be pursued and commercialized, on the other, their experience, product maturity, and diversity of applications and features offers a compelling foundation upon which to explore the potential of these new technologies. “Adobe Sensei is our AI and machine learning technology, and it is for the creation and delivery of digital experiences,” says Mejia. “We are thinking about how to make it easier, better, faster for our users to do what they do best.”
“If we think about AI and machine learning, we’ve been working on this for over 10 years. Our first machine learning feature was actually red eye reduction in Photoshop,” says Mejia. Features like red eye reduction, which were magical options in photo editing apps 15 years ago, are now automated and taken for granted. We’ve moved from slow and analog, to slow and digital, to fast and digital, to now, in many cases, automagical. It’s astounding, the ways in which design has transformed over the past 30 or so years thanks to advances like these.
[Photo: by Thomas Stephan on Unsplash]
Making the Creative Process More Efficient
Adobe emphasizes that, with the artificial intelligence of Sensei, they want to minimize tedium from the practice of design. The Sensei AI can start handling some of the rote production chores that are often required of the entry level designer. Examples of this might include the unenviable task of painstakingly cutting a complex object out from a photo background, or perhaps creating a set of graphics that are formatted properly for all the various flavors of social media. With the help of such AI-driven features, instead of requiring a detail-oriented designer to execute these tasks, the software frees them to spend their time doing other things. “I have not met a designer who got it right the first time,” says Mejia. “They like to mull over it and think about it. What AI allows you to do is … figure out what you’re trying to do faster.”
This acceleration gives the designer the opportunity to solve more problems, or even to move up the value chain and solve different classes of problems. “One of the things that AI is going to do is make those beginning careers a lot more interesting,” says Mejia. AI will transform the profession of design more radically than almost any other creative profession. Designers will be able to focus on higher order context — for example solving problems related to systems, as opposed to the pixels that frequently come to mind when people think about “designers” in a generic way.
Finding the Hidden Patterns
The ways in which Adobe is now making it easier for designers to find the right stock photography is a good micro case study of how AI and machine learning can enhance creative work — both in design as well as other professions. Stock photography has long been “tagged” with metadata to make it easier for people to search. But with AI’s ability to read and interpret images, it can tag them with a far broader range of terms that enable precision in the search. “One of the things that we’re doing with Sensei is making it easy for people to talk about what they want and what they intend, and find it in ways that feel intuitive,” says Mejia. Between the granular specificity of how the photos can be classified, and then the more advanced and user-friendly tools like a visual search, the machine learning layer is imbuing intelligence into the process that both saves time and adds quality and value to the final creative choices.
“We’re finding patterns, we’re taking that computer vision and helping to enrich the information so that, for the user, it just feels like magic,” says Mejia. “So you can do things like, for example, search for ‘family’, and then drag-and-drop in the color scheme you’re trying to match. And now you have all of the stock photography of families that are in the color scheme.”
Or, perhaps you might feed the software images that you capture out in the real-world, which hold creative inspiration for you and — buoyed by the machine’s ability to read images and the voluminous quantity of existing and well-vetted images already in the system — the AI can connect you to more and better assets. Features like this will enable the designer to remain in the space of problem solving, inspiration seeking, and solution articulation, with many fewer manual interim steps.
Automation and Setting the Creative Intent
These AI solutions will tangibly impact the designer’s workflow, especially when it comes to scaling work across the all-too-many forms of digital media. “The designer sets the creative intent. They know what they want to do. They may even create the first one,” says Mejia. “But even if you think about production or post production, creating the variations, that’s where AI is very powerful. … Being able to do that quickly means that you can spend more time on it, even with the increased pressure on production.”
[Photo: by Lubos Volkov for UX Store on Unsplash]
“So, as we think about automation, the things that are easiest to leverage AI for are things that are easily defined and repetitive,” says Mejia. “So, if you think about producing images for an e-commerce site, for example, where you may be getting lots of different inputs, but you want it to have a certain aesthetic, maybe even the same background color, et cetera, things that typically these days are a list of steps — that’s going to be something that you can leverage AI for. Even in the creation, it’s going to make some of the more individual pieces easier to do. So you get the creative more in a creative director mode, rather than a production assistant.”
For creative professionals, and designers in particular, much of the impact AI will have in these early days will be production-oriented. This ranges from tasks like design production, to writing formulaic stock blurbs, to automating the most common customer service interactions. While this, in the most literal sense, will make some jobs obsolete, it does not necessarily mean that the opportunities for a particular profession, such as design jobs, will become more limited. Looking back over the recent history of design, even as software automated certain elements and made tasks that once took hours or days now take minutes or hours, the net result was that more and more design blossomed in the world. Companies like Apple, Inc. showed the incredible impact that better design can have on corporate profits and success, making design a higher strategic priority for companies hoping to compete. As a consequence the profession of design flourished. There is no reason to believe these new advances will be any different, even as specific tasks and jobs become obsolete. The opportunities will change. Designers just need to be thinking ahead and preparing for it.
The Human in the Loop
Artificial intelligence is great within a narrow and limited context, and terrible in a larger or more multi-faceted context — especially a context as complex, dynamic, and always in motion like humanity, society, and culture. The core insights of designers, that lead to new trends and fashions on one hand, and the precisely brilliant solution for a particularly complex situation on the other, will not be achieved by machines anytime soon. This is where designers should focus their time, effort, and expertise as they prepare to succeed in the future — not on the making and fabricating, which the machines will increasingly do better than us, but on the problem solving activities like research, analysis, strategy, and solutions. What are some tasks that will remain for people to complete in the years and even decades ahead? “Breaking the mold, seeing new patterns, that spark, that’s uniquely human,” says Mejia.
“We commissioned some research and did over 100 qualitative interviews where we were asking designers what they were excited about, where they felt they could use [AI tools], what they were concerned about,” says Mejia.
“The biggest fear [they had is] that it would become homogeneous. If everyone had access to these kinds of tools, would we start to see art look all the same?”
While that is a legitimate concern, we’ve already seen this in the analog world as one trend follows the next. Despite this, new styles and ideas find their market, and the aesthetic world continues to evolve. Ultimately, it is the market as much as the creator that leads us toward or away from uniformity.
Democratizing Design
In our conversation, Mejia surfaced an important business model evolution: Individual design consultants, or agency businesses, currently rely on workflows where their ability to implement simple, production-oriented updates to existing assets, keep them busy and billing and financially viable. This will increasingly be compromised and perhaps not be replaced by internal designers and creatives, but rather by people we would not typically imagine being responsible for overseeing asset reuse and formatting — people like marketers and product managers.
“I think that AI is a necessary and important tool that will help us cope and meet the demands of all of this content velocity, of being on 24/7, of having to stand out from the crowd,” says Mejia. “It will also democratize some of the creativity, which means that not every single change will need to go back to designers.”
AI-driven software will easily replace the most formulaic and procedural tasks which, in the process, allow business people to implement what were once thought of as creative processes, requiring the creative individual to make their money doing more actually creative things. The structure of Adobe’s software already speaks to this blending of tasks and roles now in the present and, increasingly, in the future. The broad suite of Adobe tools includes modules for not just designers but a wider array of creative and business-focused professionals. Operating in the same ecosystem, an ecosystem that is increasingly integrated and bringing one another into each other’s workflow and tools, portends a fluid future where the machines tend to the tedium and the creatives take on challenges increasingly in the front-end of the creationary process.
We are heading into a period where creativity will become increasingly accessible and democratized. Adobe’s software is bringing in non-designers and putting them into a workflow with designers, as well. And, over time, more and more of the more simple or routine design tasks will become automated and not need the designer to complete. Adobe’s AI provides another big proof point for the ways the current and next generation of machine learning tools will make it easier for non-specialists — people not trained in things like writing or design — to begin professionalizing their ability to communicate or create in ways that previously would have required calling in a writer or designer. This opens up a plethora of possibilities for companies and individuals alike, which we look forward to sharing in future installments of this series.
Creative Next is a podcast exploring the impact of AI-driven automation on the lives of creative workers, people like writers, researchers, artists, designers, engineers, and entrepreneurs. This article accompanies Season 3, Episode 3 — AI-Powered Design Tools.
