Good for the Many or Best for the Few? A Dilemma in the Design of Algorithmic Advice
Oded Nov, professor of Technology Management and Innovation led this research.
Applications in a range of domains, including route planning and well-being, offer advice based on the social information available in prior users' aggregated activity.
A team of collaborators, including Graham Dove, a research assistant professor at NYU CUSP; Martina Balestra, a post-doc at NYU CUSP; and Devin Mann of NYU Grossman School of Medicine, considered a dilemma in the design of AI-based advice applications. Using an experiment, they studied whether, when designing these applications, it is better to offer:
• Goal-Directed advice: advice that is more likely to result in users who adhere to it achieving their goal, but which users are less likely to adopt and adhere to, or
• Adoption-Directed advice: advice that is more likely to be adopted and adhered to by a greater number of users, but which is less likely to result in a user fully achieving their goal
To explore this question the collaborators conducted an online experiment undertaken in four advice domains (financial investment, making healthier lifestyle choices, route planning, training for a 5k run), with three user types, and across two levels of uncertainty. Their findings suggest a preference for advice favoring individual goal attainment over higher user adoption rates, albeit with significant variation across advice domains; and discuss their design implications. The research is published in the Proceedings of the ACM on Human-Computer Interaction.