Your doctor’s AI notetaker may be making things up, Ontario audit finds (arstechnica.com)

An audit by the auditor general of Ontario has raised significant concerns about the reliability of AI medical scribes, which have been increasingly adopted by overworked physicians to streamline patient documentation. The report found that all 20 AI scribe vendors tested, approved by the Ontario government, exhibited issues with accuracy, including hallucinations of non-existent patient details and failures to capture critical mental health information. These errors could lead to inadequate or harmful treatment plans, potentially jeopardizing patient health outcomes. The findings highlight the urgent need for improved oversight and validation of AI technologies in healthcare settings. With many healthcare providers relying on AI for efficiency, the audit underscores the importance of ensuring that these tools deliver accurate, reliable support rather than introducing additional risks. This incident serves as a crucial reminder for the AI/ML community to focus on enhancing the robustness of AI systems, particularly in high-stakes environments like medical care, where precision is paramount.