Building Trust in AI: A Multifaceted Approach - SRI Toronto
CONFERENCESESSION EN ANGLAIS
11/02/2025 | 11h00 - 12h15 | Salle Cavaillès
Informations
Trust is paramount for the responsible deployment of AI in all sectors. Aligned with the AI Action Summit Trust in AI "solutions" and "standards" pillars, this roundtable will delve into the contextual and institutional dimensions of trust in AI systems. Understanding how trust is built between groups of people, institutions and technologies is essential for thinking about how AI systems can be built to reliably address human needs while mitigating risks. A forthcoming paper from the Schwartz Reisman Institute on human-machine learning interaction highlights that trust in AI systems is multifaceted, encompassing not only the perceived legitimacy of system outputs but also the trustworthiness of the developer, the deploying institution, and the administrators managing its use. Drawing on these findings and others, the panel will explore the contextual and institutional nature of trust-building in AI systems across these key areas:
Integrating trust considerations into technical and policy solutions for AI development, which includes mitigating risks, ensuring equitable access to benefits, and enhancing the trustworthiness of AI systems within specific geographic contexts.
Developing and implementing standards for evaluating AI-related risks and verifying institutional compliance with evolving trust frameworks.
Exploring pathways towards an international convergence of standards for AI risk assessment that effectively address the multifaceted nature of trust in AI.
Based on these foci, the roundtable participants will offer short position statements on multifaceted approach to building trust in AI.
SRI will outline key findings from its white paper on trust in human-machine learning interactions, highlighting the importance of a diverse disciplinary context in developing a grounding definition of trust that can support the technical design and governance of AI.
CIGI will address positions and coalitions toward international governance standards, focusing on major risks and the necessity of building trustworthy global AI.
The Forum on Information and Democracy will discuss the development and deployment of trustworthy AI systems in the information and communication space, aiming to position AI as a common good, based on a recent policy framework.
UNESCO Digital Policies and Digital Transformation will outline issues of AI trustworthiness in the context of human rights and equity.
Sciences Po will present sociotechnological perspectives on trusted AI, focusing on how social scientists—particularly in economics and political science—approach the integration of trustworthy in AI within their research frameworks.
