The Ordre des traducteurs, terminologues et interprètes agréés du Québec (OTTIAQ) has published a position paper on artificial intelligence (AI) in translation, interpreting, and terminology. It underscores both AI’s benefits and risks, with the goal of protecting the public through informed, controlled use.
Professional translators have been including neural machine translation (NMT) in their workflows since 2016, albeit with the necessary processes in place to correct the machine’s inevitable mistakes. But the rise of generative AI tools like ChatGPT presents translators with a new set of challenges, since these tools, while effective, can produce inaccurate translations and even make things up. As a result, OTTIAQ recommends involving a human expert in any translation process, especially for complex and sensitive documents where translation errors can have serious consequences.
There are also privacy, data security, and quality issues associated with this new technology. With free AI tools, there is no guarantee that the texts fed into them will be kept confidential. Indeed, they can also be used to produce other translations or train the AI. In other words, these tools don’t comply with Québec or Canadian laws on data protection and use servers in other countries. Moreover, since they’re trained on online content, language models can reflect cultural biases and compromise language quality, since they can’t work consciously or interpret information the way humans can.
OTTIAQ therefore supports the use of generative AI translation tools, but not without rigorous oversight and strict controls by professionals. We ask our members to include a human-led quality control process in any translation workflow to comply with the Order’s Code of Ethics and protect the public. We also encourage the public to use these tools with caution and to work with a certified translator whenever a reliable translation is needed.
Soumaya Boumazza, OTTIAQ