Provide AI-powered feedback on clinician interview performance

I’ve been using an AI scribe for nearly a year. In the past couple of weeks, I’ve discovered that after an interview with a new patient or a complex follow-up, I can ask the AI scribe to “give me feedback.” It will then analyze my interview, commenting on strengths and areas for improvement. This has been a game changer for me. As an attending in practice for over 10 years, I have very few opportunities for direct observation and feedback on my patient interviews. I frequently see patients with nurses, social workers, and medical students, but due to power differentials, they are not in a position to offer feedback if I make minor mistakes. I can also consult with colleagues on my team, but they were not in the room and are influenced by my own framing and biases. Some examples of feedback it has given me so far: \- It recognized that I was providing too much reassurance to a patient with OCD and suggested I be more mindful of that. \- It suggested that I be more systematic in developing safety plans with suicidal patients, something I’ve become a bit lax about over the years. I’ve since returned to doing this more consistently. \- It helped me be more nuanced in documenting a patient who wanted to die by stopping medical care, distinguishing this from suicidal ideation. It’s been like having a colleague available for immediate consultation. Of course, some of its feedback isn’t helpful (e.g., suggesting I expand on areas I intentionally left out due to low yield or time constraints), but I can simply ignore that. What are your thoughts on this? Have you experimented with anything similar?”