< Back to Cross-Project Collaboration

Fairness in Social Robotics:
Gender as a Case Study for Developing a Multidisciplinary Framework for Social Robotics and Socio-Legalstudies of AI

2021-02-01 – 2021-05-01

Short Description

Much recent effort aims to promote the development of trustworthy AI. Among key requirements for trustworthy AI are diversity, non-discrimination, and fairness, meaning that the development must support inclusion and diversity in the entire cycle of an AI system, from design to deployment. For example, when building an AI system using datasets for training purposes, attention to gender diversity is necessary in order to avoid gender biases, but also to be able to handle notions of gender or other social categories as the system interacts with human subjects, and even adapts in this interaction. This project focuses on gender diversity as a case study, bringing together a team of social roboticists (Uppsala Social Robotics Lab, Uppsala University) and experts of socio-legal studies (LundUniversity). The aim is to define a multidisciplinary framework to study how to design and develop fair and trustworthy AI for social robotics.

Outcome and Results

At the beginning of 2021, Professor Ginevra Castellano, from Uppsala University, and Associate Professor Stefan Larsson, from Lund University, set out to understand how Sociology of Law could help conceptualise gender fairness in the design of social robots (link: https://wasp-hs.org/blogposts/gender-fairness-in-socio-legal-robotics/). Laetitia Tanqueray,recruited at Lund University, helped as project assistant, funded by the project, to map out the two disciplines.

A pilot study focused on peripartum depression, a depression that can occur at any time during the conception to up to 1 or 2 years after giving birth. We conducted a study in the Swedish context, to understand what norms emanate from the medical institution and pregnant/newly-mothers. This was investigated by creating a vignette and coding a robot specifically for this project, completed by Tanqueray in collaboration with the Uppsala Social Robotics Lab, with Tobiaz Paulsson and Mengyu Zhong. This allowed the team to create a realistic robot and scenario in the peripartum depression setting. Afterwards, expert interviews were conducted with gender studies scholars and peripartum depression experts. The findings pointed to the need for interdisciplinary research to understand the complexities and sensitivities of such a context. This includes the role of the robot itself and whether it would help challenge gender fairness.

During the project’s period we also organised virtual lab visit of the Uppsala Social Robotics Lab in 2021 and a visit in person by Larsson and Tanqueray in 2022.

The collaboration resulted in three publications:

Tanqueray, L., Paulsson, T., Zhong, M., Larsson, S., & Castellano, G. (2022). Gender Fairness in Social Robotics: Exploring a Future Care of Peripartum Depression. 2022 ACM/IEEE International Conference on Human-Robot Interaction.

Tanqueray, L., Castellano, G., and Larsson, S. (2021). A Preliminary Case Study on Gender Norms in Robot-Assisted Diagnosis of Perinatal Depression: A Socio-Legal HRI Perspective. GenR workshop, RO-MAN 2021.

Larsson, s., Liinason, M., Tanqueray, L., and Castellano, G. (2022) Towards a Socio-Legal Robotics: A Theoretical Account on Gender and the Social Mirror Effect. In preparation for submission to the International Journal of Social Robotics.

Another output of the collaboration are two submitted joint research proposals with Castellano and Larsson as PIs (VR and Forte). In the context of the Forte proposal, we established connections with additional organisations, including the Jämställdhetsmyndigheten, and the Uppsala and Herlsinborg municipalities, and AI Sweden.

Responsible

Ginevra Castellano
Professor, Uppsala University
ginevra.castellano@it.uu.se

WASP-HS Project: The Ethics and Social Consequences of AI & Caring Robots. Learning Trust, Empathy and Accountability

Stefan Larsson
Associate Professor, Lund University
stefan.larsson@lth.lu.se

WASP-HS Project: AI Transparency and Consumer Trust