If the words can be translated instantly, the question gets more interesting, not less. A look at the difference between transmitting speech and managing meaning, risk, emotion, privacy and accountability in real time.
THE SHORT VERSION
- Real-time machine translation is moving from demo to everyday product.
- Think of AI interpretation as a risk spectrum, not one replacement category.
- Privacy and accountability matter as much as accuracy when live speech is involved.
- Interpreting is not ending; it is becoming more precise about where it creates value.
In June 2026, Google introduced Gemini 3.5 Live Translate with near-real-time speech-to-speech translation across more than 70 languages. In September, Google expanded its live-translation experience again, including longer background sessions and easier listening on mobile devices.
The direction is clear: real-time machine translation is moving from demo to everyday product.
For interpreters, that can sound like the beginning of an uncomfortable question. If a phone can translate the conversation instantly, what is an interpreter for now?
The answer depends on the conversation.
For low-risk situations, the technology is already useful. Travel, simple directions, casual multilingual conversations, routine service interactions, where speed and accessibility can matter more than perfect nuance. Making basic communication available to more people is a genuine benefit.
Consider a medical appointment. The speaker hesitates before answering a question. A family member interrupts. A doctor uses a term that has a precise clinical meaning but also needs a patient-friendly explanation. Someone says “yes” politely while their tone suggests they did not understand.
Which part of that conversation is “the translation”?
Or consider a live business negotiation. A sentence may need to preserve not only its content but its level of certainty, politeness and strategic ambiguity. A joke from the stage may need to land instantly. A speaker may abandon their prepared slides and improvise.
The interpreter is working with language, but also with intent, context, timing and responsibility.
This is why the most useful way to think about AI interpretation is not as one replacement category. It is a risk spectrum. At one end are high-volume, low-risk interactions where machine interpretation can make communication dramatically more accessible. At the other are medical, legal, diplomatic, regulatory and high-profile situations where an error has consequences beyond awkwardness.
“Interpreting has never been only about moving words quickly.”
Nimdzi’s 2026 research reflects that split. AI dubbing and real-time voice translation are expanding, especially in lower-risk environments, while high-stakes interpreting remains an area where expert human oversight continues to matter. Remote interpreting is also still a significant growth area for major language providers.
The technology may also change what human interpreters do before and after the live moment.
AI can support terminology preparation, transcripts, glossaries and post-event content. Real-time tools can help with accessibility or provide a fallback for languages where human coverage is difficult to source immediately. The strongest workflows may combine both rather than force a single answer for every situation.
There is another issue that buyers cannot ignore: privacy. Live speech can contain health information, legal strategy, internal business data or personal details. “Does it translate accurately?” is only one question. “Where does the audio go?” is another. And then there is accountability.
When the stakes are high, people want to know who is responsible for the interpretation. A professional interpreter can ask for clarification, signal a problem, follow a code of practice and make a judgment call when the source itself is unclear. A system can be designed with safeguards, but the governance around it becomes part of the service.
So what is an interpreter for in 2026?
Increasingly, not every sentence.
But absolutely the situations where meaning cannot be separated from risk, context, human behavior and trust.
Real-time translation is here. That does not end interpreting. It forces us to become much more precise about where interpreting creates value and where technology can responsibly take the first turn.
Language Shift by Transinity: AI, localization, careers and culture from inside the language industry. Follow Transinity on LinkedIn and join the conversation.