Post by Phrase

Your translation quality scores look fine so why is your content still underperforming in market? Most AI translation quality tools are built on MQM, a framework designed for human reviewers. The latest independent research on AI translation evaluation confirms that reliable error detection at the content level remains an unsolved problem, and MQM-based automated LQA inherits that directly. As Craig Stewart, Director of AI Research at Phrase advises: "A generic catalogue of error types was never the right unit of evaluation. It's standing in for what this particular customer means by quality." The question worth asking is what your content is for, and whether it did that job in market. Read Craig's take on why the industry is asking the wrong question and how Phrase is thinking about quality evaluation differently. Link to full article in comments