UXa Masterclass 2018: Enhancing Machine Learning with Better User Experience
This year, the Masterclass was held in Milan, Italy, just around the corner from the famous Duomo Cathedral. In contrast to the beautiful archaic city-scape, the talks focused on the ever-changing landscape of technology and evolving UX practices.
The theme of the conference was “UX Beyond the Screen” and encompassed several keynote speakers, workshops, and two streams of presentations centered on user-centered AI and customer experience design. Hiroto and I gave a talk in the user-centered AI stream titled “Enhancing Machine Learning with Better User Experience.”
With the recent surge in popularity of machine learning systems, many companies have begun to include these systems as features of their end-user products. But how well these features create valuable user experiences remains a pertinent question. As with any new technology, the initial implementations may not be fully meeting the needs and expectations of intended users.
In the first half of our talk, we gave a brief introduction to machine learning and discussed its usage in recent years. Examples included product recommendations and digital assistants, such as Siri and Alexa. We also identified some of the issues that still hamper these systems despite this technology’s growing usage base and its proliferation in everyday products. For example, recommendation systems sometimes offer impractical or otherwise inexplicable recommendations and often lack the ability for users to give feedback. We also addressed privacy concerns regarding personal data and the absence of user control.
While the reasons behind these issues vary, we cited two cases. One was the fear of missing out, which can cause companies to prematurely develop machine learning products or features that users find unnecessary. We also referenced the technical limitations of the systems themselves such as the need for professionals to be extremely mindful of the input data, how easy it is for these systems to overlook certain aspects of user needs, as well as transparency concerns.
Finally, we advocated that in order to better address these issues moving forward, it is necessary for us to constantly reevaluate our approaches. In the past, UX and machine learning professionals may have had minimal contact, but these days they should be working together more closely.
In the second half of our talk, we discussed some ways this collaboration may be achieved through the lens of organizational functions. We touched on two areas of potential improvement including how organizations create internal teams with a focus on getting the right talent and facilitating appropriate team frameworks. We also covered how machine learning systems are implemented within organizations with both UX and machine learning professionals being involved in the initial stages. We also expressed the importance of finding ways to foster cooperation through the entire development cycle.
At the Masterclass, we also heard a number of great talks with topics ranging from big data implementation to service design. However, one interesting and recurrent theme was voice UI design, such as in smart speakers like Amazon Alexa and Google Assistant.
Despite their rising global popularity, the machine learning behind these voice UI systems have yet to achieve the ability to understand things like vague or nuanced language. In order for users to naturally interact with these systems there are still many issues that need to be solved.
While some best practices do exist – such as having the system recall information from previous conversations as well as making sure users understand when the system is expecting input – when we consider the likelihood that the number of systems implementing voice UI will increase, industry professionals will need to be more proactive. In other words, we need to be considering how to best define interfaces, build prototypes, and then test them.
In general, one of the key takeaways of the conference was that utilizing new technologies like machine learning will push the boundaries of what is possible by creating novel interfaces and functions, such as interactive voice UI. However, new territory also means a host of new challenges. As UX professionals, we need to realize these issues and limitations from the early stages. That way, we are prepared to provide optimal UX regardless of the technological innovation. In the end, we shape the future of great user experiences not just by providing new gadgets but by asking the right questions and solving the right problems.