Implementing AI Transparency: Stakeholders Perspectives and Novel UX Designs
Mentor: Sara Colombo, sara.colombo@tudelft.nl
Problem: Users often lack visibility into how AI systems arrive at their decisions, the dataset they use, or their potential ethical risks. The lack of transparency practices has negative effects not only on user rights and privacy but also on informed, ethical decision-making, and trust. In the absence of transparency efforts, the ethical and social implications of AI use may remain inadequately addressed, potentially leading to undesirable consequences and eroding public confidence in these technologies.
Goal: The project seeks to understand the viewpoints of key stakeholder groups, e.g. the public, ethicists, and technologists, regarding the need for, and potential benefits of, informative processes to improve AI transparency for final users. It aims to produce human-centered UX proposals for implementing transparency in AI products (see e.g. [1]).
Methods: The project will employ a combination of surveys, interviews, and focus group discussions to gather opinions and insights from each stakeholder group. Ethical theories and frameworks will be used to analyze the collected data and identify common themes and key differences. The project will also generate novel UX designs that experiment with diverse approaches to implement transparency in AI-based solutions.
Impact: The project will offer valuable insights for designers, developers, policymakers, industry leaders, and ethicists, informing the development of user-focused transparency standards and regulations in the AI field. Furthermore, it will produce UX concepts that exemplify user-centered practices for enhanced AI transparency, which can lead to discussion on the adoption of similar practices in industry.
Relevance: This project is very relevant in light of the growing importance of regulatory processes to ensure transparency in AI technologies, particularly from a user-centered perspective.
[1] Luria, M. (2023, June). Co-Design Perspectives on Algorithm Transparency Reporting: Guidelines and Prototypes. In Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency (pp. 1076-1087).
