How Artificial Intelligence Will Change UX

This year, the official buzzword for every business is “artificial intelligence.” You’ll be hard-pressed to find an organization that is not assessing AI’s capabilities to automate tedious functions, enable greater accuracy and efficiencies, and deliver measurable benefits to the bottom line.

In the UX world, the impact of AI and automation is transforming the role of the designer, and presenting a new set of challenges and opportunities. Traditionally, UX teams would turn to metrics and tools such as usability tests, usage data and heat maps, to understand how to improve the functionality and effectiveness of a system. However, in the age of AI, we now have a myriad of empirical, actionable data that we’ve never been privy to before, giving us greater granularity into optimizing the user experience.

The user experience is fundamentally about one-to-one, human-to-computer interactions – one based around predicting how individuals would react to a set of structures or aesthetics. But with the advent of machine learning, businesses are adopting a more quantitative approach to UX, as more measurable data finds its way into the strategic decision making process. For example, Forrester argues that machine learning is what will ultimately boost the end user experience, while other pundits estimate that UX will overtake price and product to become the key consumer brand differentiator by 2020.

Improving UX skills

Pessimists have voiced concerns that the emergence of AI-powered automation will negatively impact UX designer jobs, thanks to AI’s ability to quickly and easily create countless variations of a design. However, what this fails to consider is that UX teams have a strong baseline of psychological and design-thinking approaches to problem solving. Their core skills revolve around having a deep understanding of the product’s functionalities, predicting how the users will interact with the solution, then serving as the “bridge” with strategic design thinking.

The need to correlate the two concepts won’t change for the UX team, but what’s shifting is the how of designing a product. As opposed to an approach centered on a checklist or list of deliverables, UX teams can scrutinize the sheer volume of information to make data-informed decisions. As users increasingly expect a more personalized experience, UX teams will rely less on applying a broad-brush approach in assuming the demands of the user, and will need to learn new skills from other fields to keep up with current requirements.

For example, while designers are detail-oriented, they have not been accustomed to scrutinizing mountains of data – more typical to the data science field – to figure out the full picture from the information provided. As such, it will be critical for designers to understand the basic concepts surrounding machine learning to spot patterns in user behaviors. Based on this understanding, they can then effectively work with data analysts to create compelling applications and websites.

Why AI beats A/B testing

Additionally, it’ll become more important to consider how to tailor the product for an entire group of users. In the past, it was common to A/B test products to determine the best option to create a system or website based on the majority vote. The obvious problem with this binary approach was that it completely discounted everyone else’s input, resulting in a design that failed to communicate most effectively with every user.

Tapping into AI tools, we can eliminate this one-sided approach to provide tailored experiences to everybody. We are already seeing this level of personalization in a number of real-world examples – Spotify and Netflix provide recommended playlists, while advertising and ecommerce focus on targeting from user preferences. (Related: “How machine learning fuels your Netflix addiction.”).

The next phase is moving to broader customization within the enterprise. Imagine, for example, someone using an enterprise application – such as a CRM system or HR solution. The interface tends to be the same for every user, with a base expectation of the person’s knowledge and familiarity in using the software. Using AI tools, however, organizations will be able to drill down to the individual level of the typical behaviors and pitfalls of using a system – and with these insights, better design the product so that it can contextually engage and guide the user based on their desired course of action. For example, the system could recognize that a person logging into a system needs to complete a sales entry, and could walk the user through how to accurately complete the task, but adjust the guidance on his or her previous behaviors.

Artificial intelligence and UX: design decisions

The good news is that companies already have the information available to customize the user experience, and AI will simply enable businesses to effectively conduct quantitative usability testing. Information that UI can easily extrapolate include:

• User characteristics, such as a person’s location and job title

• Device from which the user is accessing the information – PC, Mac, tablet or mobile

• Session time and length, including the time of the day the user is accessing the application

• Source from where the user arrived at the application or website – such as a Google search, typing in a URL, or clicking on a link from another website

• User flows within the application

• Drop rates from this flow when using the application

• Screen recordings of any drops from user flows, to help analyze user behavior  

• Total number of users, unique visitors and sessions

It’s critical that the right set of solutions for UX teams – including guidance, tools, validations and mechanisms – can surface the right tools for an individual at the right time. This level of information is being acquired via AI. For example, if the data shows that a user – or a segment of users – previously failed to complete a specific form in an enterprise application, we automatically surface the guidance and tips at specific failure points to support this task completion.

Designing systems truly for the user

Artificial intelligence algorithms help make our jobs simpler and easier – and by taking advantage of AI to learn user behaviors, UX teams can quickly solve design problems to create models based on user preferences, and develop more personalized applications.

As businesses and users become more inclined to greater customization, it’s vital for designers to become data-savvy, and dive headfirst into scrutinizing the endless possibilities in engaging with any system. Ultimately, it’s this transition that is delivering a strategic business impact – whether it be through lower IT helpdesk queries from users, greater efficiency and accuracy, and overall productivity improvements.

Learn more:

Real-time UX: why application performance should focus on users

Machine learning