What Should Be Considered to Support well-being with AI: Considerations Based on Responsible Research and Innovation

TL;DR

The paper addresses designing AI-enabled wellness tools that respect individual autonomy while promoting well-being, tackling the tension between generalized wellbeing definitions and personal preferences. It uses a running app example to illustrate potential perverse effects and advocates for user-driven control over how healthy behaviors are promoted. It adopts Responsible Research and Innovation (RRI) to structure design around inclusion, anticipation, reflexivity, and responsiveness, detailing practical governance mechanisms for post-deployment adaptation. The work emphasizes diverse stakeholder involvement and adaptable policies to reduce harm and enable autonomous human specification of AI behavior in dynamic social contexts.

Abstract

To achieve people's well-being with AI systems, we should enable each user to be guided to a healthier lifestyle in a way that is appropriate for her or him. However, there is a dilemma between general well-being as defined in academic and medical discussions and the autonomy users should have when deciding how to promote their well-being. In this position paper, we discuss the difficulty for AI application developers to fully consider in the design phase what might happen to the user, taking an example of a running application (app). We sort out the required factors to enable AI apps that support well-being to address the dilemma between unilaterally defined well-being and human autonomy based on the four dimensions required for responsible innovation: inclusion, anticipation, reflexivity, and responsiveness.