When Fluent Communication Is Not Reliable Communication
A recent study of AI-assisted interpreting in traditional Chinese medicine raises a problem that extends well beyond healthcare: communication can appear successful even when important information has been lost or changed.
Published in Frontiers in Public Health on 1 October 2026, the qualitative study by Bowen Geng and Shan Zhu interviewed 21 TCM practitioners in China who had experience treating international patients through informal interpreters supported by general-purpose AI tools.
The practitioners reported instances of shortened explanations, conceptual misunderstandings, and uncertainty about what information had actually reached the other party. These are reported experiences, not independently verified measures of translation accuracy or clinical outcomes.
For organizations working across languages, the findings raise an important distinction.
Fluent communication is not necessarily reliable communication.
What happens to meaning?
Consider a technical concept that has no straightforward equivalent in another language.
A translator may choose a familiar expression that sounds natural and appears precise. But if that expression belongs to a different system of knowledge, it may introduce assumptions that the original speaker did not intend.
In the study, one practitioner described the TCM concept 血虚 being rendered as anemia. Another recalled 上火 being translated as inflammation.
Both English expressions belong to established biomedical terminology. Their familiarity may help a listener form an initial interpretation, but it can also obscure distinctions between the concepts being discussed.
The same challenge arises in business, law, education, and technical communication.
A term may appear to have an obvious equivalent while carrying different implications for authority, responsibility, procedure, or commitment.
A fluent translation can conceal those differences.
Who controls the information?
The study also identified another concern. Practitioners reported that interpreters sometimes condensed lengthy explanations, omitted information, or answered questions on patients’ behalf.
In those situations, the communication problem was not limited to terminology. It involved decisions about what information passed between the participants.
Organizations face similar risks whenever communication depends on intermediaries.
A meeting summary may preserve the decisions while omitting the reservations expressed during discussion.
A translated report may communicate the main recommendation while losing important qualifications.
A customer-service exchange may accurately capture the customer’s request while missing the concern behind it.
None of these outcomes necessarily produces awkward or obviously incorrect language.
The communication may appear complete.
Reliability requires more than fluency
AI tools can make multilingual communication faster and more accessible. They can also help people communicate in situations where professional language support is unavailable.
But linguistic fluency is only one part of communication quality.
The larger question is whether the process has preserved the information and distinctions that the participants need in order to understand one another.
That requires attention to three things: the conceptual assumptions carried by translated terminology, the decisions made about what information is transmitted, and the opportunities participants have to identify or clarify misunderstandings.
The study does not establish that AI-assisted interpreting is inherently less reliable than other approaches. Its contribution is to make visible the difficulties practitioners reported within one particular communication arrangement.
For organizations, the wider lesson is straightforward.
The fact that a conversation proceeds smoothly does not establish that everyone has understood the same thing.
Reliable multilingual communication depends not only on producing language that sounds right, but on preserving the relationships and distinctions through which meaning is understood.
Research reference: Bowen Geng and Shan Zhu, “TCM practitioners’ perspectives on AI-assisted ad hoc interpreting in China: a qualitative study”, Frontiers in Public Health, 1 October 2026.
Continue exploring
Fluent language can make communication feel more settled than it really is. The harder work begins when we ask what has happened to the meaning underneath: what was preserved, what was interpreted, and what another person may have understood differently.
Meaning Alignment explores what happens when professional communication is linguistically correct but meaning remains unstable. Through workplace examples and practical exercises, it develops the interpretive skills needed to recognize hidden assumptions, judge nuance, understand professional and relational signals, and identify where meaning may begin to diverge across languages and cultures.
Comments