Presentation
Whose Wisdom? Human Biases for Decision Support System Sources
SessionPoster Session 2
DescriptionHumans make quick judgments about decision support systems (DSS). These impact their use of the system, so it is critical to identify what influences these initial judgements. This experiment investigated if these opinions are affected when users only know the DSS source — machine learning algorithm (ML) or wisdom of the crowd (WoC) — and size of its underlying data — small, medium, or large training data or crowd size used to obtain aggregate estimates, respectively. 78 participants chose between pairs of the DSS sources (e.g., medium ML vs large WoC), then completed a visual search task with recommendations from the chosen support. Results indicated no preference between WoC and ML, but a clear preference for larger DSS data sizes. After six trials of assistance, participants were calibrated in their estimates of DSS reliability. The findings suggest users may have an innate understanding of the benefits of increased training data or crowd size in improving recommendation performance, and are able to make quick, accurate judgements about the DSS. This work highlights the potential of considering the wisdom of the crowd as a viable DSS, especially in tasks where it might outperform machine learning.
Event Type
Poster
TimeWednesday, October 15th5:30pm - 6:30pm CDT
LocationRiverside East
Similar Presentations