The launch of Meta’s new artificial intelligence model, capable of instantly translating speech-to-speech or text-to-speech communications in up to 101 languages, presents a technologically advanced and probably very useful tool.

There are plenty of artificial intelligence models that facilitate automated translation and interpretation, which have been available for quite some time. These tools are available to professionals in different fields, although they do not all work in the same way: some are open access, some are proprietary, and some have been created by companies themselves for internal use, among other variants.

Language plays a fundamental role in a variety of contexts, more than we can imagine. Translating for a medical conference, a film, a high-level meeting, an investment seminar, an interview with a producer or a film debate is not the same thing.

It is therefore crucial to understand the context and purpose for which we use these translation/interpreting models, and they should be tailored to each individual case.

What level of quality do we expect to obtain from these tools in order to consider them minimally acceptable from a professional point of view?

For the time being, these technologies can only be a valuable support tool in our profession, and we will be keeping an eye on their evolution in the future.