Intelligence, evolved: strategizing, designing, and developing for AI
Artificial intelligence. In its most simple terms, it’s a computer program or software that develops the ability to reason by analyzing data patterns in the form of text, image, voice, and video. Research into this field began over 60 years ago during a workshop at Dartmouth, spearheaded by John McCarthy and his team of computer scientists.
However, thanks to cloud computing and more sophisticated neural networks, we’ve seen significant advances in this technology over the last decade. Today, you’ll find its initial application everywhere, from healthcare, to autonomous cars, image recognition, and voice-powered assistants.
With artificial intelligence making its way into the mainstream, many designers and developers find their work is shifting away from apps and websites to more complex data-driven products, such as virtual assistants, voice-based applications, and personalized experiences developed with deep user behavior and data.
A few weeks ago, we had the pleasure of partnering with the NYC Chapter of IxDA, to bring together a series of experts from across the AI spectrum. Hosted by Smart’s VP of Design, Kelly Clark, we welcomed speakers representing UX, content strategy, and technology respectively:
- Dominic Poon, Associate Director at IBM Interactive Experience
- Elizabeth Holli Wood, Content Strategy Consultant
- Patrick Way, Senior Software Engineer at Intersection
Each came equipped with a different perspective on what it’s like to work with artificial intelligence, what they’ve learned, and the key takeaways for our audience to incorporate into any future AI endeavors.
Machine learning is a whole new ball game
As with other emerging technologies, creating big data experiences is new territory for many designers and developers. Everyone “designing for firsts” is doing so with without precedent, input, or guidance, but as Dominic points out, this evens the experimental playing field exponentially.
There are dozens of elements to working with AI that we’ve never needed account for with product design in the past. The first of these is a shift in traditional game theory, which studies the conflict and cooperation between intelligent rational decision-makers. No longer are we designing for one player, but two: The user, and their artificial teammate.
A lack of traditional interfaces means that user experience and satisfaction will not only be different, but will need to be measured, tested, and improved upon in new ways. We need to hone in and focus on what makes interacting with AI unique from other technologies. Tone, personality, context, intent, national language content curation: all of these are aspects we can take advantage of to truly differentiate it from experiences like websites and mobile apps.
Above all, Dominic reminds us it’s worth remembering that while machine learning as we know it today may be obsolete in five years time, what it offers will still be relevant. “Technologies can go away.” he maintains, “What’s going to stick around is the capability.”
Dominic then left us with his five IBM-rooted design mantras:
- Know your data. The AI will analyze your data, and churn back the results.
- Design for honesty. No matter how popular AI is, users need you designing solutions to their problems.
- Show, don’t just do. AI presents a new type of data. Illustrate its difference, and what makes it unique.
- Deliver insights. Not just visualizations.
- In service of people, never in lieu of them. Designers influence how people think. We will define how the world views AI.
You don’t have control over what’s going to happen next all the time. The AI does. So you create holistic experiences.
Meaningful data is the key to unlocking intelligent products
In addition to the information we feed it, AI have the ability to listen, learn, and create new data sets of their own, giving us an entirely new design language to work with. For the first time in history, we no longer have full control over the experiences we’re building.
But in actuality, this is not as scary as Hollywood has led us to believe: In addition to facial recognition from Snapchat and voice assistants like Siri, there are dozens of positive applications for big data. Just last week, Andrew Ng announced his latest creation, Woebot: A chatbot that can help those suffering from depression.
Using data responsibly is our ticket to changing the world for the better. With artificial intelligence, the golden rule is that you’ll get out what you put in. In the case of Tay, a Twitter chatbot designed and built by Microsoft to emulate a 19 year old girl, the data she fed off of from internet trolls “attacking” her caused her to react with inflammatory messages of her own. Even though the experiment was pulled after just sixteen hours, muses Elizabeth, it taught an invaluable lesson: “What’s feeding an intelligent system is what you’ve decided is meaningful, and what you’re putting in it.”
The corpus an AI is working with and creating itself requires constant vigilance: It should be monitored, scrubbed, and maintained in order to allow the program to be the best version of itself.
“When designing for an AI experience,” Dominic mentions, “you don’t have control over what’s going to happen next all the time. The AI does. So you create holistic experiences.” By doing this, we have the wherewithal help close in on the the notion of globally ubiquitous computing.
What’s feeding an intelligent system is what you’ve decided is meaningful, and what you’re putting in it.
Designing for the human element has never been more important
Our relationship and how we interact with technology is rapidly changing, and AI is no exception. From the data that machine learning creates for us comes an entirely new design language: One based on empathy, emotion, and trust.
The foundation of this lies at the intersection of content strategy and data, the lines of which are increasingly blurring. In order to create artificial intelligence that humans will develop a meaningful relationship with, we must focus on letting shine facets that appeal to their emotions in ways their other devices can’t: Personality, brand, and the “who, what, when, where, why” of products matter more than ever.
“The branding element is all about the verbal identity.” Elizabeth tells us, “What is the brand voice of the product? In the the case of AI, that’s increasingly an actual voice. You’re having a conversation with a box in your living room, and we’ve determined that’s okay, so now we have to determine how we want that relationship to be.”
Three takeaways Elizabeth gave us to consider:
- Content does not live in a vacuum. It weaves through everything, just like every technology.
- Create tech with it, not against it. Take content, UX, and tech into consideration together, not individually.
- Think beyond empathy as usual. Consider who you’re making it for, who you’re leaving out, and what implication that has.
AI is evolving the way we build products
All three of our speakers touched on the way teams working on these new technologies are adapting their methodologies, starting with their structure. Each were in agreement that an independent, siloed, and deeper focus on skills and expertise, rather than a generalist understanding, is required from designers, yet they should remain highly collaborative with developers.
Due to the evolving, volatile nature of these machines, content, UX, and technology should all be concepted together, then receive the development validation it needs before taking the first steps of being brought to life.
Rapidly prototyping new technologies isn’t easy, particularly when the experience you’re simulating might not be anything tangible at all, like a conversation with Google Home. Traditional methods of team collaboration and illustrating to one another what’s being built, such as storyboards, wireframes, presentations, and screen-based artifacts, fall short.
The solution? Scripts.
“Script writing” asserts Patrick, “is an established discipline with rules, guidelines, and practices. Because of this, it’s fantastic for defining experiences that don’t necessarily fit into other design artifacts.” Through prototyping and scripting combined, we’re able to closer portray the concept of interacting with voice-based assistants, chatbots, and other AI technologies.
Harnessing big data for a better future
As with any new technology, there’s a huge learning curve and plenty of trial and error before we we really find the sweet spot. The good news is that we’re not alone in designing for this “first”, and thanks to experts like Dominic, Elizabeth, and Patrick, the future of machine learning is looking very different than films like Terminator would have led us to believe.
We’ve learned that the data that you choose to implement is just as much of a UX and technological decision as it is a content decision. Thanks to AI’s ability to create and evolve its own data sets, we need to look at data as an enabler of brand and personality, not just a foundation or defined factor of what makes up a product.
But above all, we’ve learned that the goal of AI is to be unique. To customize itself to its user. To find new ways of helping those who interact with it, and in turn, serve to benefit those who don’t. By paying attention to, harnessing, and embracing the nuances of a properly and responsibly crafted artificial intelligence experience, we can, as Dominic mentioned, create a more ubiquitous world.
Thanks to our event partner IxDA, and their NYC Chapter lead, Peter March. To learn more about the digital product capabilities of Smart Design, reach out to our team for a chat by emailing email@example.com.