Start Date
11-8-2016
Description
Based on big data, decisions can increasingly be drawn from data-driven analytics and algorithmic decision support. However, it remains unclear whether recommendations issued by computer algorithms are equally accepted by individuals as human advices. This is particularly intriguing given that big data entails various forms of ambiguous decision situations in which individuals cannot assess the underlying database or possible consequences. We conceptually introduce ambiguity in the light of big data and conduct an experiment to identify whether individuals prefer algorithmic to human recommendations and how outcome and data ambiguity affect individuals’ adoption behavior of algorithmic decision support. We find a preference for human recommendations, independently from the level of inherent ambiguity. However, areas that are more data-driven have a higher potential to overcome resistance to algorithmic decision support. Our results imply that developers of algorithmic decision support should provide high levels of transparency in areas which currently lack support.
Recommended Citation
Fuchs, Christoph; Matt, Christian; Hess, Thomas; and Hoerndlein, Christian, "Human vs. Algorithmic Recommendations in Big Data and the Role of Ambiguity" (2016). AMCIS 2016 Proceedings. 5.
https://aisel.aisnet.org/amcis2016/Adoption/Presentations/5
Human vs. Algorithmic Recommendations in Big Data and the Role of Ambiguity
Based on big data, decisions can increasingly be drawn from data-driven analytics and algorithmic decision support. However, it remains unclear whether recommendations issued by computer algorithms are equally accepted by individuals as human advices. This is particularly intriguing given that big data entails various forms of ambiguous decision situations in which individuals cannot assess the underlying database or possible consequences. We conceptually introduce ambiguity in the light of big data and conduct an experiment to identify whether individuals prefer algorithmic to human recommendations and how outcome and data ambiguity affect individuals’ adoption behavior of algorithmic decision support. We find a preference for human recommendations, independently from the level of inherent ambiguity. However, areas that are more data-driven have a higher potential to overcome resistance to algorithmic decision support. Our results imply that developers of algorithmic decision support should provide high levels of transparency in areas which currently lack support.