Suggesting that LLM should be renamed to LCM ASAP Ever since all the major AI models have been going all out in the Coding and Agent tracks, coupled with the existence of the so-called 'mysterious distillation phenomenon,' things have gotten super weird: 1. The programming capabilities of major AI models have skyrocketed 2. Everyone's abilities are becoming increasingly similar 3. (Chinese) language proficiency is plummeting So, what was initially called LLM (Large Language Model) now seems more fitting to be called: Large Coding Model. This rapid decline in language ability? My gut feeling is that it started with GPT-5. It can’t speak human anymore, just spouting corporate jargon from big tech. Then it spread like wildfire. Now even DeepSeek struggles with Chinese, and today I tested Grok 4.6—its Chinese output was equally unbearable. Honestly, calling the old GPT a bad apple that spoiled the bunch doesn’t feel like an exaggeration. Everyone’s distilling from each other back and forth, and you can barely tell the difference anymore. Oh, and even Opus, which used to be a Chinese language expert, is now on a downward spiral. And what about Gemini? Its programming skills are getting crushed, but at least its Chinese language ability is holding steady. Is it impossible to excel at both coding and language expression? Save Chinese, please. If not, in the future, AI will consume all of humanity’s linguistic data, then churn out polluted AI jargon as new training data. I can’t even imagine—if this cycle continues, human language might eventually devolve into gibberish that only AI can understand.