Data Subjects’ Conceptualizations of and Attitudes Toward Automatic Emotion Recognition-Enabled Wellbeing Interventions on Social Media
Published in Proceedings of the ACM on Human-Computer Interaction, 2021
Recommended citation: Kat Roemmich and Nazanin Andalibi. 2021. Data Subjects’ Conceptualizations of and Attitudes Toward Automatic Emotion Recognition-Enabled Wellbeing Interventions on Social Media. Proc. ACM Hum.-Comput. Interact. 5, CSCW2, Article 308 (October 2021), 34 pages. https://doi.org/10.1145/3476049 http://kroemmich.github.io/files/3476049-a.pdf
We argue that to increase the trustworthiness of automatic ER-enabled wellbeing interventions on social media, companies that deploy them would need to at least fulfill requirements that preemptively protect individuals from the vast harms it presents, take measures to attenuate harms, and align with data subjects’ development and design requirements. These requirements include high computational accuracy, contextual sensitivity, positive outcome guarantees, individual controls, external regulation, and meaningful consent over being subject to automatic ER-enabled wellbeing interventions. We conclude with a message of caution and restraint about the use of automatic ER-enabled wellbeing interventions on social media in the US, based on its current regulatory landscape and social context.
Received an Honorable Mention for Best Paper Award (top 5%) at CSCW 2021.
Recommended citation: Kat Roemmich and Nazanin Andalibi. 2021. Data Subjects’ Conceptualizations of and Attitudes Toward Automatic Emotion Recognition-Enabled Wellbeing Interventions on Social Media. Proc. ACM Hum.-Comput. Interact. 5, CSCW2, Article 308 (October 2021), 34 pages. https://doi.org/10.1145/3476049