AI continues to gain traction across healthcare, but that adoption remains cautious... especially in patient-facing clinical conversations. A new study published this month in NEJM Catalyst: Innovations in Care Delivery, "Artificial Intelligence for Language Access in Surgical Care: Patient Preferences and an Implementation Framework," takes a closer look at where AI-based interpreting tools may fit and where professional human interpreters remain essential.
Researchers at Mass General Brigham evaluated two modalities in standardized postoperative consultation scenarios with twenty-three Spanish-speaking surgical patients reviewed both approaches and shared their preferences across:
Rather than framing AI and human interpreters as competing solutions, participants consistently positioned them as complementary tools. In routine or time-sensitive exchanges, speed and privacy tended to matter more often. In emotionally complex or high-stakes discussions, empathy, accuracy, and cultural nuance became critical.
Patients did not select a single “best” option. Instead, they emphasized flexibility and choice — describing interpreter preferences as dependent on the clinical moment, emotional stakes, and type of conversation taking place and matching the right modality to the right moment.
Participants consistently favored Jeenie interpreting for emotionally sensitive or high-stakes postoperative discussions — such as conversations involving fear, uncertainty, or complications. Human interpreters were associated with empathy, cultural nuance, and reassurance — particularly among patients with low English acculturation.
“If I’m scared or it’s something delicate, I prefer a person.”
AI tools were appreciated for speed and perceived privacy in more straightforward or time-sensitive exchanges. At the same time, patients expressed concerns about dialect recognition and contextual fidelity.
“I can use it wherever I am, at any time.” and “With AI I feel like no one else is listening.”
but also... “AI may not understand some words we use in my country.”
The authors write, “Our findings suggest that patients likely do not perceive artificial intelligence as a replacement for human interpreters.” Instead, the study supports a hybrid model that strategically integrates AI and human interpreters based on urgency, sensitivity, and patient preference.
In a moment when AI adoption is accelerating across many industries, this independent, peer-reviewed research reinforces an important truth: professional human interpreters remain essential in clinical conversations — and thoughtful AI + Human integration is what patients actually want.