Episode #16
Clinical Writing, Generative AI, and the Bias Problem: A Conversation with Annie Holland
In this episode of On Record, hosts Carmen Haak and Kyle Shelstad speak with Annie Holland, a faculty member in the Communication Sciences and Disorders department at the MGH Institute of Health Professions, part of the Mass General Brigham Health System and one of the top-rated Speech-Language Pathology programs in the country. Annie is also the inaugural recipient of the CAPCSD-IVS Clinical Education Seed Grant. Annie shares how her path from pediatric clinical practice, including work in neonatal intensive care and school systems, led her to graduate education, where she now teaches master's-level SLP students with a strong emphasis on clinical writing. She discusses her published research on biased language in clinical documentation, which had students identify and rewrite problematic language in sample reports before and after a targeted training module. The conversation covers specific examples of where bias tends to appear in SLP writing, including assumptions around gender and cultural identity, and why getting that language right matters for families, patients, and the broader care team. From there, Annie talks through her current IVS-funded research project, which uses a multi-institution team of experts to analyze AI-generated speech-language evaluation reports for biased and subjective language. The goal is to understand what these platforms are actually producing, across multiple tools, and in conditions that reflect how students and practicing clinicians are using them right now. She offers practical guidance for programs considering AI documentation tools, including the importance of prompt quality, the environmental cost of excessive trial-and-error prompting, and the need to go in with a clear plan rather than hoping for good output. One of the more unexpected findings she shares: the strongest resistance to AI in her department has not come from faculty. It has come from students, many of whom have well-considered concerns about how these tools are being used in clinical practice. Annie and her colleagues have had to rethink how they introduce AI into the curriculum as a result. The episode closes with how video recording connects all of this work at MGH IHP's on-campus clinic, where live-streamed sessions give students room to practice independently, recorded sessions anchor clinical writing exercises and AI output review, and end-of-semester video shares let the entire cohort learn from one another's patient work. Annie can be reached on LinkedIn or via email at AK.Holland@mghihp.edu. Her published research on biased language in clinical writing is available open access in Teaching and Learning in Communication Sciences and Disorders. More information on the CAPCSD-IVS Clinical Education Seed Grant is available at capcsd.org/scholarships.
