If there's one problem that LLMs have solved, it's language. While an LLM may hallucinate, it does so in grammatically correct English sentences. Additionally, even the local version of gemma-4-26B can seamlessly switch between languages in the midst of a conversation while maintaining context. That's perhaps the most exciting part for me: We have a bonafide universal translator (that's Star Trek territory) and people seem more focused on its factual accuracy.