In response to your article ( ‘Being human helps’: despite rise of AI is there still hope for Europe’s translators?, 8 May ), I work freelance for a company that produces memoirs for its customers. I used to interview, then write. Now, I interview, a large language model writes, and I am paid half of my previous fee to edit the result. It takes as long to edit the AI-generated text as it used to take me to write the memoir. There are several reasons for this. First, the AI content is tedious and homogeneous. I could ignore that problem, but I have a relationship with the often vulnerable people I interview. Second (and yes, I know that AI is constantly improving in this regard), it produces rubbish if an interviewee is less than coherent. Third, I am required by my employer to “check for accuracy”, which the company presents as a quick additional task, but which is in fact impossible without a full cross-reference of interviews against the draft.…