Over the past several years, UX has evolved from a relatively siloed discipline into a foundational approach to digital marketing and design. And as emerging technologies like artificial intelligence (AI) generate new forms of user interactions, UX thinking is taking on fresh and vital roles.

UX-related questions are virtually endless in this fast-developing space, especially in emerging areas such as voice recognition that leave the familiarity of the screen behind. With no template for how these experiences should look, sound and feel, UX Researcher and Filterati Hannah Sherwood explains, it’s the principles of UX that will pave this uncharted path with usability, accessibility and effective storytelling.

Currently conducting user research for Microsoft’s Cognitive Services, Hannah is deep in the nitty-gritty issues tech leaders are facing as they shape these experiences: practical challenges that don’t typically make their way into the sweeping, hype-filled headlines surrounding AI. She explains that her research is driving home the same concept again and again — one she believes is one of the biggest keys to grounding AI in real user needs.

In this interview, Hannah explains that AI is most effective when companies break from the trend of pursuing it as one big, groundbreaking solution. Instead, she argues, it’s better approached as a collection of highly specific tools — many already more familiar than most of us think — that they can use to craft focused and truly user-centric experiences.

Read on for Hannah’s insights into what user experience really means in the context of AI, and how slowing down to unpack AI into its distinct parts can help companies better leverage its potential.

Q: How do does UX thinking fit into the creation of AI experiences?

Any time there’s a new technology, people get very excited about what it can accomplish. That’s definitely the case with AI, and understandably so. However, it’s easy to get caught up in buzzwords and lose focus on the real purpose of these experiences. There’s an urge to create AI tools that can “do it all,” such as a chatbot who can talk about anything. That approach can lead to products that sound exciting but don’t necessarily target a specific need, and that set up unrealistic expectations for users.

UX is helping AI technology concentrate on the core tasks at hand: who the user is, what they need to accomplish, and what will help them complete those steps as quickly and easily as possible. Often, that means guiding companies away from what’s “cool” to what actually works best for their customers and their business goals.

When designing AI experiences, I think it’s incredibly important to take a step back and assess the reasoning behind our decisions. UX thinking can help us get to that big picture.

 

h22.jpg

Hannah Sherwood

UX Researcher and Filterati Hannah Sherwood helps leading brands understand their users. Need help in this area? Let us know.

Q: Your recent research has focused on conversational interfaces. What makes these experiences unique?

With conversational interfaces, questions that have always been at the heart of UX become even more apparent. We’re not just talking on pixels on a screen; we’re examining all the different pieces that come together to create that holistic experience. Storytelling has always been part of the UX toolkit, but that skill set becomes even more important as we think about these new types of interactions.

Typically, people think about UX in terms the interface of a website, app or product, and how we use those visual cues to navigate from point A to point B. When we start to think about things like voice interfaces or chatbots, suddenly there’s not really a screen anymore. Take the interface out of the way, and what are we left with? This forces us to think more fundamentally about what those experiences should look like.

Sometimes, this means determining that a conversational interface isn’t the best fit for the problem at hand. Say, for example, I’m an airline that wants to build a chatbot to help customers plan their trips. UX research might reveal that it would save the user more time and steps to forego the conversational interface and use a different machine learning tool (or tools). For instance, a better solution might be an algorithm that automatically suggests flights based on past behaviors. Whether the answer is a conversational interface or something else, UX principles help keep these decisions more grounded in actual usability.

Q: What is an example of a usability challenge UX experts are working to solve in conversational interfaces?

There are many UX challenges to tackle in this area, and most of them arise from the fact that people tend to overestimate these tools’ capabilities. Users often attempt to interact with voice and chat software the same way they would interact with another human. But so much of human communication is unspoken: though the technology is rapidly improving, a product like Alexa simply can’t make the kinds of inferences and connections that a person can.

When a machine can’t process what the user is trying to communicate, it’s easy to get stuck in a frustrating a loop of “I didn’t understand that, try again.” That scenario may leave the user with a negative impression of the interface and even the brand as a whole.

When the AI piece gets stuck, how can we make it a better experience for the user? What will that conversational flow look like? Can the chatbot provide more detailed instructions, or offer options based on the parts it did understand? As UX Researchers and Designers, we can help give users a clearer understanding the tool’s capabilities and limitations, setting up “guardrails” that stop them from hitting these blocks.

Q: Are there any misconceptions you believe are hindering companies’ AI efforts?

On a fundamental level, I think there’s a widespread lack of understanding about what AI really is. I’ve seen even technology leaders talk about AI in vague or inaccurate terms. Extend that to the general public, and it can start to feel like just a buzzword.

I think a big part of the problem is that people often think about AI in a very science fiction sense. They picture a machine that that can seamlessly interpret and respond to all kinds of visual and contextual information, just like we can. That’s what people in the field refer to as “general AI”, and it’s an incredibly hard problem, one that we’re unlikely to solve in our lifetimes.

I don’t believe “AI” is the necessarily the most meaningful term to use, because it’s not about “intelligence” in the way that sci-fi often conjures. It’s much more about automation and processing information, and we’re doing a lot with it already. When people think about AI, they’re not thinking about the machine learning applications most of us are already using.

Contrary to a lot of the buzz, I think the most significant impacts of AI are actually going to be very subtle: small moments of automation that will make our day-to-day work smoother and more efficient, hopefully giving us more time to tackle more interesting problems. By staying grounded in that reality, we can better understand, and more creatively utilize, the tools that are available to us. I think the main takeaway is that AI isn’t really one thing: it’s combinations of smaller parts that can be pieced together to solve a wide variety of problems.

Q: What steps can companies take to best leverage these opportunities?

Start exploring the wide range of AI “puzzle pieces” that are out there and how they can target your users’ specific needs. Think about the kinds of things you’re doing now that would benefit from automation, and then explore what kinds of tools could automate them. Microsoft’s Cognitive Services suite, for example, has 30 services and APIs that cover everything from image and facial recognition to text and even sentiment analysis. Keep in mind that if the technology you want doesn’t exist this very second, the pace of change in this industry is so rapid that it may well be on the market soon.

And of course, keep user-focused thinking at the core of your AI research and decisions. UX skill sets will continue to be as important — if not more important — than they are today in shaping these tools. If you truly understand your users, and can anticipate and address their needs, then you’ve already won.

Comment