Abstract

Human-algorithm interaction emerges as the new frontier of studies involving user experience (UX) and information architecture. By examining sensitive aspects of the users’ interaction with machine learning (ML) algorithms, this research focused on content consumption and interaction with Spotify’s streaming media recommendation system. One-on-one interviews were applied to Brazilian university students and it revealed that they had not precisely understood how the system works and had not developed complete mental models concerning how data is tracked and handled to generate recommendations. Transparency and explainability are requirements for ML systems development as these systems are similar to black boxes, providing no clarity about how they work. Systems should tell users what implicit and explicit interaction data are collected to create recommendation lists. Another point concerns data privacy: users have revealed a suspicion of what might happen to their data and have exposed some “conspiracy theories.” Answers suggested that there may be communication failures, and UX designers and information architects should make evident how the system tracks and processes user interaction data. This chapter also relates interconnections between machine learning and information architecture, and frictions between artificial intelligence and UX.