Scottish AI Alliance Supports Ground-Breaking AI Safety Initiative: The PHAWM Project

The Scottish AI Alliance is proud to support the Participatory Harm Auditing Workbenches and Methodologies (PHAWM) Project, an initiative designed to enhance the trustworthiness and safety of AI technologies.

About The Project

AI auditing has become essential in the current landscape of evolving regulatory frameworks. However, traditional AI audits have been inconsistent and typically conducted solely by experts. The PHAWM Project, a collaboration between seven academic institutions and 23 partner organizations, seeks to revolutionise this process by introducing participatory AI auditing.

A New Approach to AI Auditing

The PHAWM Project introduces the concept of participatory AI auditing, where diverse stakeholders, including domain experts, regulators, decision subjects, and end-users, can audit AI systems either individually or collectively. This inclusive approach ensures a broader perspective on potential harms and enhances the reliability of AI systems.

The project’s research is grounded in four key use cases: Health, Media Content, Cultural Heritage, and Collaborative Content Generation. These areas have been selected to demonstrate the wide-ranging applicability of the participatory auditing approach.

Tools and Training for Effective Audits

To facilitate these audits, the PHAWM Project will develop workbenches—tools designed to help stakeholders assess the quality and potential harms of AI systems. Additionally, the project will establish methodologies to guide how, when, and who conducts these audits. Training programs will be provided to equip stakeholders with the necessary skills for participatory auditing, and efforts will be made to create a certification framework for AI solutions.

Collaboration and Expertise

The Scottish AI Alliance joins a prestigious group of partners in this initiative, including The Data Lab, NHS NSS, Open Data Institute, Research Data Scotland, and Cybersalon.

The project benefits from the expertise of an advisory board along with distinguished academics from institutions like the University of Cambridge, University of Oxford, and more.

A Step Towards Safer AI

The Scottish AI Alliance is committed to fostering safe and reliable AI development. By supporting the PHAWM Project, we are taking a significant step towards ensuring that AI technologies are developed and deployed in ways that are trustworthy, ethical, and inclusive to all.

Steph Wright, Head of the Scottish AI Alliance had this to say about the project:

“This proposal addresses the challenges of deploying predictive and generative AI in a way which is aligned with our vision for responsible AI in Scotland. Its emphasis on investigating auditing across a diverse range of contexts and its aim to improve equality, diversity and inclusion in AI deployment match well with Scotland’s AI Strategy. We are particularly excited by the novel nature of participatory auditing, as we believe that through this approach new technology development will be informed by those who will be most impacted by predictive and generative AI.”

For more information on the PHAWM Project, visit the PHAWM website.

Previous
Previous

Unlocking Public Sector Data: Introducing Research Data Scotland’s New Researcher Access Service

Next
Next

New digital service for public sector data open for use