AI translation tools are part of the current landscape of technical translation. They are useful, they are limited, and they raise legitimate questions about confidentiality, quality, and professional responsibility. My position on the matter is straightforward and is based on two principles.
I do not run your content through public machine translation engines. Whether your documents contain proprietary specifications, pricing structures, pre-release product information, or internal procedural content, they stay in a controlled environment.
This is a deliberate choice with practical consequences. Public AI translation services typically retain rights over the content submitted to them, use it to improve their models, and offer no enforceable guarantees of confidentiality. For technical content tied to a specific industry, a specific client, or a specific product line, this is a material risk that most organizations do not need to take.
The decision to use AI on your content belongs to you, not to your translator. If your organization has established its own AI-assisted workflow, with controlled tools and clear confidentiality boundaries, I work within it. If your organization prefers a non-AI workflow, I work within that as well.
If your team produces an initial draft using AI tools — DeepL, in-house models, or other systems — I provide the certified review, correction, and OTTIAQ-recognized sign-off your final materials need. This is increasingly common for organizations that have integrated AI into their internal documentation pipeline and need a certified professional to validate the output before publication.
A note on AI-assisted technical translation. Modern technical translation rarely consists of typing a French version from scratch while reading the English. The work today combines human expertise with a set of tools: machine translation as a starting point on some projects, translation memories that surface previously approved phrasing, terminology databases that maintain consistency across documents, and AI-based reference systems that can be consulted on terminology questions. The substance of the work — producing a French text that reads as if it had been written in French and that meets publication standards in a regulated industry — has not changed. The means of getting there have.
What separates a strong result from a weak one in this kind of workflow is not the choice of tools but the experience of the person using them. The same machine translation output produces a clean starting point for one translator and a tangle to unwind for another, depending on how it is read, corrected, and reworked. Fifteen years of work in industrial and engineering documentation is what allows me to use these tools as instruments rather than crutches.
If your organization prefers a fresh translation from the source, that option is available as well. The appropriate approach depends on your content, your timeline, and your compliance requirements, and I will recommend the one that best serves the project after reviewing a sample.
AI does not replace the professional judgment required to bring a translation to publication standards in a regulated industry. That judgment, and the responsibility that comes with certifying the result, is the substance of my work. The tools change; the underlying expectation that a certified translator stands behind the final text does not.