If you’ve found yourself saying “Hey, Siri, what’s artificial intelligence?” into your phone, you’re definitely not the only one to experience that high-tech irony. From predictive searches in Google to voice-powered virtual assistants, artificial intelligence has increasingly become part of our everyday lives. But despite its widespread proliferation and buzzword status, mystery around what AI is and how it’s integrated in apps, like Spark, remains.
So, what is AI anyway?
Broadly defined, artificial intelligence is the capacity of a computer to learn and make decisions in ways that mimic human cognitive processes. To varying degrees, applications or systems that demonstrate the ability to reason, discern meaning, make generalizations, or improve choices based on past experience or user interactions are utilizing AI. Scientists and engineers can imbue software with AI in a variety of ways. To take examples from Adobe products, this can range from coding up simple rules, like keeping faces from being obscured by text, to complicated machine learning models, which can answer questions like “what makes a quality image?”
Anyone who’s explored the world of visual communication even a little knows content creation is filled with both judgment calls requiring a human eye and expertise, (“Is this readable? Do the image and message relate to each other?”) as well as tedious, repetitive tasks (“Ok, everything looks good, now I need to resize it for nine different social media surfaces.”) Adobe has set out to ease some of this cognitive and manual load for humans through artificial intelligence. It’s even instituted a virtual company-wide engine that unites the intelligent features across Adobe products, called Adobe Sensei. This framework ensures Adobe scientists across teams are tackling common problems in a shared AI-powered way towards a unified effort to improve the design and delivery of digital experiences.
At Spark, our scientists are on a mission to rid the design process of as much tedium and guess work as possible, so you can focus on creativity and message. As Joe Shankar, director of engineering, puts it,
“the driving question of our problem solving is always ‘How can we help you get results without you having to sweat the details?’”
We use AI and machine learning to bake design intelligence right into the product so when you’re creating content in Spark it’s like working with a trusty graphic designer who knows all the design principles and always has time to check your alignment or help you resize a graphic. After a stringent internal application process to prove our technology is worthy of Sensei distinction, Adobe Spark officially joined the ranks of this prestigious artificial intelligence and machine learning standard bearer. Or, as the team likes to joke on Slack, we entered the dojo.
Spark, powered by Adobe Sensei
The best AI is subtle and stealth—ninja-like, if you will. So if you’ve used Spark Post before, you’ve already experienced Spark’s design intelligence through a number of powerful features. But there’s some real black belt-level AI in Spark Post’s Magic Layout feature, which makes creating standout collages and sophisticated graphics faster and easier—for anyone—than ever before.
Behind the Curtain
Today we’re revealing a few tricks behind Spark’s Magic Layout with the help of Peter O’Donovan, senior computer scientist, who along with Joe, leads Spark Post’s design intelligence technology.
Usually creating sophisticated graphics or collages with multiple design elements is a pain. According to Peter, most free-form design tools have very little intelligence or algorithms built in and therefore require a user to crop, position, and resize elements manually, relying on their own design knowledge and expertise. As such, the novice designer or average communicator is up against two hurdles: navigating the manual tools and employing graphic design principles they may not know.
Built-In Graphic Design IQ
Spark Post’s Magic Layout sets you up for success quickly by employing a content-aware layout system to suggest intelligent cropping, layout positions, image zooming, and text placement based on the content you choose for your design. It makes educated guesses for you to get you almost all the way to a finished product. For instance, if you choose one tall photo and one wide photo and add a text element, Spark Post will automatically place your content in a layout with one vertical and one wide cell. Even more amazingly, it will position an editable text box where the system detects there’s relative empty space and common design rules predict it will be most readable. The system also attempts to determine which part of the image is the most important or the most interesting and positions the image accordingly.
These intelligent suggestions mean you can skip the tedium and head straight to the creative refining stage—where a human eye (and tricks of the trade and style!) take the graphic to the next level. But just because Spark has suggested something “smart” to get you started, doesn’t mean the AI is over.
Getting to Wow Faster
Spark also lets a user explore countless options with just a tap of the finger. And this is where seasoned designers can appreciate Spark’s simple approach to designing too. Any pro designer knows you gotta try a lot of things to land on the winning combination, but without machine learning, that’s a lot of tedious work. “Spark’s AI is aimed at allowing users to traverse the design space quickly,” says Peter. “Why shouldn’t a user be able to try 20 layout combinations in under minute?”
While the user can cycle through several intelligent options, the system also allows them to make tweaks to the layout elements directly and have other elements respond automatically. So say you add a new text element or a new image, the layout suggestions update automatically, immediately learning from your input.
This smart, rapid exploration allows you to reach creative solutions that would have taken much longer with traditional manual tools. Or as Joe puts it, it allows you to “increase design velocity.”
But getting to an end-result quickly isn’t the only thing his team is solving for with machine learning. “We also want to make sure you’re creating something that looks good,” he says. For that, we bake in design and communication best practices into every template while our engineers build a dynamic system that learns from the growing body of designs our community creates everyday.
Land on a beautiful finished product in minutes by using our starting points, or dive into the design canvas with ease to learn along with Spark.
Give Spark Post’s Magic Layout, powered by Adobe Sensei, a whirl today! The following remixable templates can help get you started. Stay tuned for more deep dives into how Spark is becoming smarter everyday!