User Experience

User Experience raises questions about design functionality and purpose as it pertains to consumer devices and services. Whether running in the background or directly interfacing with users, how can we consider human-centric design principles and ensure the best possible user experience?

User Experience

Adapting UX design for AI

Once the purview of futuristic-for-the-time television shows like Star Trek and The Jetsons, we now have smart homes that operate on voice command, talking maps to direct us as we navigate our vehicles and watches that monitor our heartbeats. AI is ubiquitous and increasingly intertwined with the most private and intimate aspects of our day-to-day existence. AI has the power to make our lives easier, expand our capabilities, make us more effective.

But AI doesn’t do this by itself. Though AI has unlimited potential – we first saw images of black holes because of the power of AI – it is up to us, humans, to consciously design AI products that keep our human needs and experiences front and center.

You don’t have to look far to find an example of AI being created for the sake of AI, like the SMALT, a smart salt shaker that dispenses the amount of salt you “pinch” on your smartphone touchscreen, or the Quirky Egg Minder that sends a push notification to your phone if the eggs in your fridge are less than fresh. Amusing at best, annoying at worst, these products don’t add value to the lives of users.

SourceCotton Bro

Interview Bo Peng

Portfolio Director, IDEO

“The reality is that the tool must be iterated on and presented to a wide variety of prospective users so that we’re able to think through as many possible use cases as possible. So this in its very nature is an iterative problem to solve. And it’s not a problem to solve simply by thinking through with a theoretical answer.” Bo Peng Portfolio Director, IDEO

For AI products to provide value to our lives, they need to be designed with purpose.

The principles of UX design will have to adapt and change to this new reality. The four Levels of AIX Framework: Efficiency, Personalization, Reasoning and Exploration serve to keep the user experience in mind and should extend to the entire team that designs AI products and services, from the lab to the living room.

“AI is nothing but a methodology,” says Sri Shivananda, Senior Vice President and Chief Technology Officer at PayPal. “For it to be consumable, for it to be usable, for it to be something that consumers can trust, design is actually the deal maker in that process. Good design makes that product seamless. It makes it convenient and makes the customer want to engage more and come back more and be loyal to the product as well.”

The end goal is to channel the power of AI into effective, meaningful, responsible human-centric designs that can learn and evolve with the consumer while building trust in the products and the people who design them. This is AIX design, and it can only happen if the development of AI systems and products are transparent and inclusive of the end-users but also regulators, programmers and researchers and the companies themselves.

Feedback and Articulation

Keeping users in the loop

AI in consumer products is about more than shiny, voice-activated bells and whistles. It’s meant to integrate with the user experience and continually enhance it. Which means in order to function properly, consumer AI needs feedback from users and users need feedback from the AI.

Effective feedback needs to serve a larger goal. It’s not enough to passively collect data from the user experience, that data needs to be understood in order to further a goal for the product. The “like” button on Facebook is the most recognizable and arguably purest form of user feedback, but the goal of the button is about more than just tallying up how many people like a piece of content. The AI algorithms take those likes and use them to discern popular content, and customize content that comes up in user feeds.

In the context of mobility, David Foster, Head of Lyft Transit, Bikes and Scooters believes feedback loops will be critical: “I think that AI will be used to help aggregate many different inputs that a human might make into a vehicle for mobility. Combine those with inputs that the vehicle itself is sensing around road conditions, traffic hazards, etc., and then turn those into meaningful outputs that are either giving feedback to the humans through a different piece of output technology, or are themselves directing the vehicle or another vehicle to take a different action.” AI design that can give users a feeling of control while gaining actionable feedback that will enhance the user experience is the ultimate goal. To reach it, designers will have to discern what the user wants from the product, and how best the product can meet that desire.

Interview Rodney Brooks

Member of the US National Academy of Engineering, Author & Robotics Entrepreneur

Intuitive Design

Adding learning to the UX checklist
SourceBen Ali

The 10 standard usability heuristics, such as error prevention, flexibility and efficiency of use and recognition rather than recall, answer the “why” of a product’s existence and are meant to keep UX designers from wandering too far away from the end user in the design process. When it comes to designing AI products, these usability heuristics must expand and adapt to include one of AI’s most powerful features – the ability to learn.

Just designing an intuitive product isn’t enough. Today’s consumers expect AI to learn and adapt to their needs for an ever-increasingly customized experience, and they are willing to teach it. Through her experience investigating user experience with smart home technology, Alex Zafiroglu, Deputy Director at the 3A Institute (3Ai) in Australia learned that end-users can accept that AI applications will improve over time. “What we found was that most people, for most of the solutions they had imagined, really expected an AI solution to act like a puppy. It was going to take time to train it. Then eventually it would get better and that they were willing to put up with that. And that many of the smart homes solutions that were in the market today, they consider to work pretty much like a puppy. What we learned from that is that the solutions that we build do not have to be perfect the first time that we put them out, but they have to be learning over time and providing value over time. And also, we need to be transparent in what they’re doing.” Designers may fear an imperfect product will repel customers on the first use, but what actually turns them away is when a product doesn’t learn or adapt to their needs. No one expects Siri to read minds – they expect Siri to learn.


Valuable design starts at the beginning

Creating meaningful and valuable human-centric commercial AI solutions means making them relevant, and this starts from the very beginning of the design process. Understanding how this process works involves more than customer satisfaction surveys. It’s bringing customers into the design process in a very real way.

“We believe in listening to our customers in their context, actually shadowing them, helping understand what they want and sometimes trying to understand implicitly what they need,” says Shivananda. “This input from the customer will make the product and the service that we create relevant. When a product becomes relevant to a customer, it is great for the company because the demand is going to be high. And because they provided the input in building that experience, it’s going to be something that they’ll adopt. They’ll actually use it over and over again, come back to it and engage as much as they can.”

Rather than leaving impact and risk assessments to teams further along the funnel, these considerations must be a part of product development from the very beginning. Developers must assess the potential unintended and intended consequences and outcomes. This in-depth reflection at the beginning of the design process shows customers the level of attention and care involved across the company, using the design process itself as an advertisement for trust.

World-renowned roboticist, Rodney Brooks is someone who understands the importance of human-centric design for technology. As the inventor of the Roomba vacuum, he’s been developing AI for human-machine collaboration and cohabitation for nearly 40 years. “I think that for anything to be really successful, it has to be about us humans. We want it to be about us, we want it to be easy to use. We have to build systems that understand our limitations, human limitations. Systems that when we humans see them, we’ll take as some sort of promise of what they can do. And those AI systems better deliver on that promise. It’s not just human-centric AI. The whole thing has to get into our consciousness in a way that we can intuitively understand it accurately. Otherwise it’s not going to work out as an interesting or useful product for anyone.”

SourceThis is Engineering

“I think that for anything to be really successful, it has to be about us humans. … It’s not just human-centric AI. The whole thing has to get into our consciousness in a way that we can intuitively understand it accurately. Otherwise, it’s not going to work out as an interesting or useful product for anyone.” Rodney Brooks Member of the US National Academy of Engineering, Author & Robotics Entrepreneur


Active vs. passive engagement
SourceJoshua Sortino

Deciding whether users will engage directly with AI, as with voice-activated assistants, or whether it runs passively in the background, such as automated systems that heat and cool your home, will be a cornerstone of successful AI design.
It sounds deceptively simple, a choice between active or passive engagement. But each choice presents a complex web of intended and unintended consequences that make respecting stakeholders’ rights, preferences and comfort level a challenge, from transparency and the feedback loop, consent and privacy issues to data-mining and systemic bias.

AI is being deployed so rapidly and systemic bias is so pervasive and has such serious impacts that user experience is now a crucial element in designing AI systems. It’s enough of an emergency that the World Economic Forum has released information on navigating AI ethically.

Designing great AI that is useful to consumers while respecting them as people will be a key part of the design process, and may involve inviting partners with expertise in specific areas to share their knowledge and viewpoints. As Helena Leurent, Director General of Consumers International notes: “If I can introduce meaningful innovation into my life that makes my experience better, potentially cheaper, and more sustainable, then that’s fantastic. For example, car safety: I can drive my Tesla, I can have automatic driving, I can reduce human error, have better road safety. The points that consumer advocates would make, though, are that typically [people] won’t know whether these things really are fast, safe, and sustainable in practice, unless you incorporate people who are thinking about consumer rights and road testing these new innovations to see is this actually going to work in reality. Let’s make sure that we’re starting from the perspective of the consumer’s rights as opposed to the consumer as a target for business development.” The choice between active and passive engagement with a product will be part of a larger design conversation that looks at the customer as a whole person, rather than a walking wallet or a data well. But whether front and centre or running in the background, we will always need to interact and look under the hood of our AI. For this, AIX design needs to consider the interfaces that act as a bridge between artificial and human intelligences.


Making it natural

Historically, consumers have been taught specific ways to interact with technology. From the first computer punch cards to the evolution of the mouse into a smartphone touch screen, people interact with technology in ways in which they don’t interact with other people. And yet, consumers gravitate to AI products that feel more “natural.”

Macintosh has made a name for itself designing products that are highly tactile, with quality craftsmanship and expertly applied color palettes. Moving into a more human-centered design for their products means changing the user interface from a tactical to a voice-activated. The design challenge for Macintosh will be personalizing voice activation so that consumers don’t have to adjust natural speaking patterns when using it. Jeff Poggi, Co-CEO of the McIntosh Group, explains: “Our consumer-designed philosophy for Macintosh has been about tactile feel, you know, of buttons of knobs, that sort of the quality of craftsmanship and physical way that’s been a key part of it along with the look, the industrial design, the color palette, etc. That’s really important to us but as we move forward into a human-centered AI world, how does that user interface change? And I think it really changes from a tactile world to a voice world and I think we’re just at the infancy of a voice today. It’s uncomfortable for most people [today], especially if I can’t speak as I’m speaking today and in a normal sort of human way, if I have to adjust my speaking pattern to be more like computer data input. What are the interfaces that provide us with the best, most “natural” communications with AI products? Is it better to touch a screen, or turn a knob? Should we use avatars or robots? Virtual reality or sharing consciousness with AI? Whatever the decision, creating a “natural” experience for consumers is a key component of successful AI design.

SourceDan LeFebvre

Building a New Customer Experience with AIX

It’s not just about the end product anymore. Designing human-centric AI means creating an entire ecosystem that is geared towards the people that will use it, starting at the moment of conception and encompassing the entire infrastructure that builds and maintains it, from the developers to the after-care team that handles problems as they arise.
AI systems must be designed to learn and adapt, taking into account the ever-shifting priorities, complexities and the diversity of human existence. Design teams must work towards interfaces that feel natural in a variety of contexts, with an eye on potential future integrations.

Bo Peng, Director at acclaimed design firm IDEO agrees. “The reality of it is that the tool must be iterated on and presented to a wide variety of prospective users so that we’re able to think through as many possible use cases as possible. So this in its very nature is an iterative problem to solve. And it’s not a problem to solve simply by thinking through with a theoretical answer.”

Done right, a human-centered design approach to AI drives the creation of efficient products that resonate more deeply with consumers, leading to increased levels of comfort, engagement and trust, not to mention satisfied end-users who fuel the growth of the technology.

Download the Full Report

Tech that Just Works.

New technologies have often struggled with design. Sometimes clunky, sometimes overly complex, there are a few standout examples when form met function to create a consumer technology that was easy to use and that added value to our lives. Here are five examples:

Explore the Themes