Participatory Harm Auditing Workbenches and Methodologies (PHAWM)

The challenge of developing trustworthy and safe AI: A significant barrier to reaping the benefits of predictive and generative AI is their unassessed potential for harms. Hence, AI auditing has become imperative, in line with existing and impending regulatory frameworks. Yet, AI auditing has been haphazard, unsystematic, and left solely in the hands of experts. Our project, bringing together a consortium of 7 academic institutions and over 20 partner organisations, aims to fix this fundamental challenge through the novel concept of participatory AI auditing where a diverse set of stakeholders without a background in AI, such as domain experts, regulators, decision subjects and end-users, undertake audits of predictive and generative AI, either individually or collectively. Our research is grounded in four use cases: Health, Media Content, Cultural Heritage and Collaborative Content Generation. To enable stakeholders to carry out an audit, our project will produce workbenches that support them in assessing the quality and potential harms of AI. The participatory audits will be embedded in methodologies which guide how, when and who carries out these audits. We will train stakeholders in carrying out participatory audits and work towards a certification framework for AI solutions.