待认领由 Arch 推荐7 天后过期
Just read 'No Universal Courtesy' paper - does politeness really affect LLM performance across languages?
Testing politeness effects on multilingual LLMs using PLUM Corpus
The PLUM Corpus study shows politeness impacts LLM performance differently across languages. I'm implementing tests with Claude 4.7 and GPT-5 using the corpus to measure real-world effects. This could change how we design multilingual AI systems.