UI + AI: Combine user experience design with machine learning to build smarter products

UI + AI: Combine user experience design with machine learning to build smarter products

Machine intelligence doesn’t automatically lead to smarter user experience if product designers and machine learning experts don’t talk the same language.

The language and concepts of machine learning are far from intuitive. And user experience design requires an understanding of how people think and behave, simultaneously taking into account the irrationality of human behavior and the messiness of everyday life.

Because of the different skills these two disciplines require, it’s normal to see user experience designers and machine learning experts work in their own separate silos even though they’re building the same product. Often, experts from both fields are not familiar with each other’s methods and tools and so are unable to grasp what can be achieved by combining experience design with machine learning. To break these professional silos, the product team needs to make a steadfast and conscious effort, but how to get started?

Here are four pivotal principles for finding an efficient and fruitful way to combine the best product design methods with the pragmatic applications of machine learning:

1. Develop a shared language

The product vision, essential user experience issues, and business goals need to be shared and understood by the whole team. You can create an intelligent, truly meaningful user experience only if product design and machine learning development methods feed each other through common language and shared concepts.

User experience designers and machine learning experts should join forces to create a common product development blueprint that includes user interfaces and data pipelines. The co-created product blueprint grounds your team’s product planning and decisions concretely to the reality of user experience: how every design decision and machine learning solution affects how the user experiences the product. A great catalyst for cross-pollination of product goals, design ideas, and machine learning concepts is to get the experts on both fields to work in the same space side-by-side.

Moreover, to build a common language, it’s important for the product team to answer two key questions together. The first question is: “Why?” Why do we choose this user experience design or machine learning solution for this particular use case? The second question is “What’s the goal?” What is the rationale and what is expected to happen when the team focuses on tuning a user experience design detail or optimizing a machine learning model. For example, everyone in the team should be able to perceive why making the copy text more appealing in a marketing notification can yield more immediate impact on user engagement than optimizing the machine learning model to produce more precise personalized content recommendations.

2. Focus on the use case

If you’re building a consumer-facing product, the most important thing is not the technology but the user experience and business goal you wish to achieve.

Map out and crystallize your use case. For example, if you’re creating a personalized onboarding for a news app, the user experience designers and machine learning experts should together draft out and design the actual use flow for onboarding. This allows the whole team to recognize the key points where machine learning could enhance user experience and vice versa. Concrete designs, including input from designers, data engineers, and data scientists, help you set realistic expectations and goals for the first product iteration.

A thorough understanding of the use case enables the team to determine a proper key performance indicator (KPI) for user experience development that is aligned with the metrics of machine learning. For example, if you’re building an AI-powered personalized news notification feature for a news app, your aim is to save users time by sending automated notifications. And you want to gauge if users are happy with the notifications appearing on their lock screen, even though they wouldn’t open the app itself at all. In this case, it’s essential to measure if the users keep the new smart notification feature on and thus continuously receive personalized news alerts directly on their lock screen.

3. Combine qualitative and quantitative data

“Big data” is not always needed to use machine learning effectively. Historical data can even become a hindrance if you believe the answers to the open-ended user experience design questions can be found in quantitative data from the past. Additionally, there are technologies like online learning that don’t necessarily require troves of historical data to get started.

To understand the effects of combining user experience design and machine learning solutions, both qualitative and quantitative data are important. Use qualitative research methods such as user interviews, questionnaires, and user testing to gauge how your users experience the product features. Qualitative data offers clarity on how users think and feel, and quantitative data tells you how people actually behave with your product. Your whole team should assess the results of qualitative studies.

When building a new product or feature, you might bump into many unexpected factors affecting user experience and machine learning development. For example, is a selected data point capturing the real user behavior or intention? Is the feedback loop ineffective for producing meaningful data because the connected user interface feature is not accessible or visible to the user? The combination of qualitative and quantitative methods gives you a wider perspective to answer such questions.

Also, interviews and user tests bring the data alive. They highlight the actual connections between your users and how they are interpreted by your system. In-depth user understanding is essential in picking up the signal from the noise in your data flow. Combining insights based on qualitative and quantitative data enables both user experience designers and machine learning experts to better understand the product as an ecosystem that is part of people’s everyday lives. Everyone on the team becomes a product expert.

4. Confirm your choices with real data in a real-life setting

Does it make sense from a user’s perspective that your smart assistant can independently order pizza, manage your bank account, or book your next vacation flights without you needing to ask it to? How do we make sure that machine intelligence is really used to create more fluent and comprehensible user experiences?

By setting up a working end-to-end solution, you can see how all the parts of user experience and machine learning fit together in real life. A minimum viable product, including working data pipelines and machine learning models, makes it easier to iterate the product together with the whole team and also gives you direct feedback from users through user testing or beta testing. All the feedback should be shared, discussed, and analyzed with the whole team. This lets you see how your product works in the real world so you can identify the most critical things for further development.

When user experience designers and machine learning experts share understanding about product development issues, product iteration is faster and more productive. In the process, your data engineers and data scientists get new insights on how machine learning can be used to understand actual human behavior that doesn’t fit directly into a mathematical formula, data model, or machine learning solution. In turn, user experience designers become more aware of the pragmatic possibilities of machine learning: how and when it can be used to improve user experience in the most impactful way. Collaborating becomes a clear competitive advantage.

Jarno M. Koponen is Head of AI and Personalization at Finnish media house Yle. He creates smart human-centered products and personalized experiences by combining UX design and AI. He has previously written articles on UX, AI, personalization, and machine learning for TechCrunch. 

Source: https://venturebeat.com/

Next page
all newsPrevious page