The latest Data Phoenix news tagged data-annotation.
Apple researchers found wide swings in how LLMs and people judged translation quality, with agreement unreliable in most settings tested.