Re-examining the role of time in human-algorithm interaction

Human-Computer Interaction;Human-Algorithm Interaction;Interaction Design;Slow Movement;Slow Algorithm

Park, Joon Sung ; Karahalios ; Karrie G

关键词: Human-Computer Interaction;    Human-Algorithm Interaction;    Interaction Design;    Slow Movement;    Slow Algorithm;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/108025/PARK-THESIS-2020.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

How fast we interact with technology has been one of the most salient topics for both the users and designers of computing systems. Experimental results since the days preceding personal computing era characterized short response time as an agent for user satisfaction, productivity, and engagement, and we have invested a significant amount of engineering effort to make our computers faster ever since. However, the role of speed in human-computer interaction is much richer and more multi-faceted than what today's culture of moving fast makes it out to be; where faster interaction offers us efficiency, slower interaction offers us a chance for reflection, serendipity, and a moment for deliberate thinking. As we transition into the days of advanced algorithms and AI where even the most consequential and often problematic judgments are made on our behalf, I see an opportunity to revisit our belief in speed, and examine what slowing down can do to empower users in the interaction between humans and algorithms. To this end, this thesis explores concrete, measurable benefits of slower interaction in improving users' assessment of an algorithm's accuracy in human-algorithm interaction. Specifically, I report a series of online and in-person between-subject user studies in which I isolate the impact of an algorithm's speed on how users incorporate the algorithm's advice when making judgments in the context of simple visual recognition tasks. I find that the participants followed good quality algorithms more and bad quality algorithms somewhat less if the response time of the algorithm is slower. Furthermore, qualitative analysis of the in-person study interviews reveals that the waiting was not time wasted, but was often used to reflect on, and think deliberately about the task and the estimation process of themselves and the algorithm, and to compare and reevaluate the two processes. Based on these findings, I outline design implications for future algorithmic systems.