Designers Will Guide AI Into the Next Design Revolution

Artificial intelligence and machine learning are changing the way products are built, but they’ll need designers’ guidance to reach their full potential.

You say, “AI-powered design,” and many designers start having nightmares of a Cylon takeover.

Since its emergence in the mid-20th century, artificial intelligence (AI) has risen and fallen in the public consciousness, typically seen through a man-versus-machine lens. However, AI has matured into a fundamental technology used in a variety of fields such as robotics, computer vision, natural language processing, and increasingly UX and product design. As a result, more and more people are becoming comfortable with the notion of a machine sidekick — even people on the creative side.

That said, AI isn’t the be-all, end-all, as I shared at Adobe MAX. It tends to perform best when trained to execute or accelerate specific tasks. Machine-learning techniques ensure that AI can continue to improve, learning more as it gains experience and is fed new information.

However, AI is not especially good at human-level cognition or “generalized” intelligence. In fact, the most promising applications of AI today have deviated from the early goals of building human-equivalent intelligence, instead focusing on opportunities where computers can perform tasks humans aren’t particularly well-suited to do (for example, making sense of large data sets).

That’s where core design and UX applications come in. Redefining AI as something that augments and amplifies humans — rather than replacing them — gives designers license to reshape how we interact with AI systems, and where AI might add value in our lives. Gone are the days of humanoid robots as the primary AI interface. If designers can envision systems that communicate, reason, remember, learn, and act, you’ll create powerful new products that are more adaptive, immersive, and better able to solve customer challenges. Yet, even a system that does one of those things really well might create groundbreaking experiences for your customers.

Six ways machine learning is transforming product design today

Certainly not every new product or experience design needs — or warrants — an AI layer. However, many could find clear-cut perks from weaving in this technology, taking the experiences into fresh new territories.

The following are examples of six core domains that get better once AI is introduced. Pairing AI with human cognition and human empathy can together create more compelling, more dynamic, and more high-value products and services, anchored in customers’ real-time wants, needs, and goals.

1. Analytics systems

AI is already widely used in analytics systems — specifically, to help companies determine how well services are performing, and help search engines decide what to display to individual customers. In both of these scenarios, trying to keep up with the data coming in and going out would be impossible as a human marketer or analyst. AI is really good at taking these types of tasks off your hands, effortlessly working through these massive data sets.

Ancestry.com is a good example of this type of AI system at work. The site has millions of scanned images, historical documents, and public records that are processed via machine learning. With these AI-powered “eyes,” century-old, handwritten birth certificates or census records can be read, categorized, and matched to potential ancestors.

From there, “hints” are delivered to Ancestry.com customers so they can quickly and easily fill in their family trees. The alternative — digging through centuries of unclassified, hard-to-read records — would likely derail countless amateur genealogists.

2. True product personalization

Machine learning is also capable of hyper-customizing experiences based on what it knows about a customer paired with trends in the general population. Netflix is a perfect example.

Netflix personalizes how the service presents different content to different customers. For example, if they know Customer A likes action movies, but Customer B tends toward romantic comedies, Netflix’s AI-powered recommendations may present different shows — or it may present the same show in completely different ways because of their expressed interests. The action seeker may see a darker thumbnail while the rom-com fan may see something lighter and more upbeat.

3. Content understanding

AI and machine-learning systems can be taught to recognize not just images, but how different images relate to one another. Many banks use this technology — customers can snap a paper check on their smartphone and deposit the funds to their account. The AI technology interprets the check, confirms the amount and the legitimacy of the check, and recognizes the account holder’s signature, then makes the deposit — no trip to the ATM needed.

Content understanding is also at play in Adobe Lightroom. Here, AI spots unique individuals in photos, proactively grouping and tagging photos accordingly. This enables automatic tagging and faster search through often unwieldy photo folders. This same technology can also interpret and tag based on other key features — whether the picture was taken near the ocean, and during the day, or outdoors, for example.

4. Manipulating content

Snapchat filters are a great example of real-time, AI-powered content manipulation. However, there are much bigger possibilities when it comes to delivering interactive experiences and accelerating content creation.

For example, this technology is being used in Adobe Photoshop’s Content-Aware Fill feature. Content-Aware Fill uses information about an image’s content and context, and when something is removed from the visual it can intelligently fill in the space. This turnkey technology is user friendly, highly intuitive, and saves designers and photographers endless time editing images.

5. Human-computer interaction

Machine learning is also unlocking new ways for people and devices to interact. Audio-based music search engine Shazam is a great example — sing a few bars and the app identifies the song instantly. More recently, researchers at UC Berkeley used machine learning to map forms and movements of professional dancers to the bodies of non-professional dancers, making even amateurs look like total pros.

These interactions aren’t just about entertainment, though. Oura Ring — a sleep and activity tracker — reads customers’ sleep quality, activity, heart rate, temperature, and more. Using this information, the ring predicts their readiness for the next day. A sleepless night or high temperature, for example, might prompt an Oura Ring recommendation to skip the gym and sleep in.

6. Advice and wayfinding

Machine learning is already being used in chatbot interactions and support scenarios where AI can guide humans, whether through the physical world — think GPS app Waze — or the digital world, such as in-product tutorials. Perhaps the most exciting aspect of this technology, though, is its potential to provide on-the-job assistance.

For example, Adobe Sensei, Adobe’s AI and machine-learning engine, is being integrated into Adobe Creative Cloud. This allows AI technology to act as a virtual studio assistant that can take on rote tasks — removing unwanted objects from photos, for example.

Becoming the guides for AI in design

It’s clear AI and machine learning can be powerful tools when applied to UX and product design. While the technology still struggles in certain advanced applications, such as original content generation and open, unrestricted conversations, it’s continually evolving and “learning.” The limit to what’s possible mostly resides in our imaginations for how to use the technology.

Designers can play a critical role in charting where we’re headed. Unique use cases and product ideas that emerge from the design community can shape the direction of new technology investments. A layer of human empathy, trust, humility, and transparency needs to come from you, your teams, and the creative community as a whole. Together, human-machine design pairings can produce unparalleled results.

Granted, this won’t be a simple swap-out for traditional processes. Each creative team needs to invest time and talent building out the right model for their unique business. This will likely involve a more iterative process, with a few missteps along the way. And, beyond that, teams will need to pay close attention to the quality of the data and be vigilant against bias.

It’s an exciting prospect for the creative community as a whole. Together, we have an opportunity to shape this next evolution of design and creativity — and that, to me, is incredibly compelling. Our job is to truly dig in and understand what AI should be through the design lens. Because, clearly, AI and machine learning aren’t simply technologies to consume or even to fear. Instead, they’re powerful tools in our creative toolbox that are increasingly driving the way we imagine, ideate, and create.

Check out the Adobe Blog for more ways to integrate AI and machine learning into your product design evolution.