A study of UX practitioners roles in designing real-world, enterprise ML systems

Opportunities for AI and machine learning (ML) are vast in current interactive systems development. However, comparatively little is known about how functionality for the system behind the interface is designed and how design methodologies such as user-centered design have influence. This research focuses on how interdisciplinary teams that include UX practitioners design real-world enterprise ML systems outside of big technology companies. We conducted a survey with product managers, and interviews with interdisciplinary teams and individual UX practitioners. The findings show that nontechnical UX practitioners are highly capable in designing AI/ML systems. In addition to applying UX and interaction design expertise to make decisions regarding functionality, they employ skills that aid collaboration across interdisciplinary teams. However, our findings suggest existing HCI design techniques such as prototyping and simulating complexity of enterprise ML systems are insufficient. We propose adaptations to design practices and conclude that some existing research should be reconsidered.