CASE AI (Crowdsourced AI Safety Evaluation: Ambient Intelligence)

Opportunity

This project will test whether crowdsourcing can effectively identify safety risks in new digital health technologies such as AI scribes and classify the levels of harm these risks pose. During the project, students, early-mid career researchers, and clinicians will be recruited to observe the use of Lyrebird’s AI scribe in simulated clinical scenarios and then identify the risks they observe in the different scenarios. This approach aims to fill the gap in practical safety guidelines for emerging digital health tools. An expert panel will review and prioritise identified risks to support safer adoption of AI in healthcare. Up to three crowdsourcing events are planned as part of this initiative.

Project Objectives

• Explore the factors that contribute to implementing a successful digital health student crowdsourcing event
• Investigate the feasibility of identifying risks and issues associated with AI scribes through university student crowdsourcing
• Determine how effectively student-identified risks associated with AI scribes can be assigned against a harm severity framework.