Wrestling with language is part of the national identity in Switzerland, where German, French and Italian are used — plus a fourth official language: Romansh – Copyright AFP Fred DUFOUR
As hospitals increasingly turn to artificial intelligence to bridge communication gaps, a new study suggests that when it comes to language access in healthcare, patients may not want a single solution—but instead, a choice.
Researchers at Mass General Brigham have evaluated the use of an AI-powered medical interpreter in surgical settings, offering one of the clearest looks yet at how such tools affect patient experience and care delivery. The findings, published in NEJM Catalyst, highlight both the promise and the limitations of AI in one of medicine’s most sensitive environments.
At the heart of the study is a simple but pressing question: can artificial intelligence improve communication between clinicians and patients who do not share a common language—without compromising trust, clarity, or emotional nuance?
To explore this, the researchers compared three approaches: traditional remote video interpretation (RVI), an AI-based real-time translation platform, and a hybrid model combining both. The AI system, developed by the company No Barrier but evaluated independently, provides instant translation across more than 40 languages.
Spanish-speaking peoples
The study focused on Spanish-speaking surgical patients at Brigham and Women’s Hospital, examining not just technical performance, but how patients felt about each method. Using a modified technology acceptance framework, researchers assessed factors such as perceived usefulness, ease of use, trust, and cultural alignment.
The results point to a nuanced conclusion. While AI delivered clear advantages in speed and convenience, it did not replace the perceived value of human interpreters.
Patients appreciated the immediacy and privacy of AI translation, particularly for straightforward or logistical communication. In contrast, human interpreters were preferred for emotionally complex or high-stakes conversations, where tone, empathy, and reassurance play a critical role.
Many participants ultimately converged on a hybrid view. As one response captured succinctly: “A combination would be ideal.”
The augmentation of existing systems
This preference for flexibility—rather than substitution—emerged as a central theme. It suggests that AI may function best not as a replacement, but as an augmentation of existing systems, allowing patients and clinicians to choose the most appropriate mode of communication depending on the context.
The implications extend beyond convenience. Language barriers are a well-documented source of inequity in healthcare, affecting everything from diagnosis to treatment adherence. By offering on-demand translation, AI tools could reduce delays and improve access—particularly in fast-paced environments like surgical care.
Yet the study also underscores that communication in medicine is not purely transactional. Trust, emotional resonance, and cultural understanding remain critical, especially when patients are facing procedures that carry risk or uncertainty.
For healthcare systems, the message is clear: effective language access may require layered solutions, combining technological efficiency with human sensitivity.
More broadly, the findings reflect a growing trend in the integration of AI into clinical workflows. Rather than a wholesale transformation, progress often comes through careful implementation, testing new tools alongside established practices to understand where they add value—and where they fall short.
As AI systems become more sophisticated, their role in patient communication is likely to expand. But this study suggests that even the most advanced algorithms cannot fully replicate the human dimensions of care.
Instead, the future may lie in designing systems that recognise this balance—using AI to enhance speed and access, while preserving the empathy and connection that patients still value most.