It has come to the attention of the Tranquality Council that the AI Safety community does not typically include translation errors in raw machine output when making an inventory of risks associated with AI. The main purpose of this survey is to add to a collection of examples so we can strengthen the argument that in some use cases, translation errors can indeed cause harm, and translation consumer labels—UVT (unverified translation) or PVT (professionally verified translation)— are needed to inform end users of how the material was produced.
We are looking for documented examples of machine output, including neural machine translation (NMT), generative AI (GenAI) translation, etc., that will help us illustrate errors that could cause harm if not corrected by a professional translator before the translation reaches the end users, be it an e-book, an article, or an advertisement.
Please, fill out the form below with at least one documented example. We need to know the field, the specific example with information related to where it was found (newspaper, magazine, website name/URL, etc.), the proposed correction, and the possible harm the error could cause.
Thank you for your contribution and you may contact Prof. Alan Melby if you have any questions. You may also click here to learn more about the Labels Project.