US models align with our "average" (western) values. If we outsource thinking by using LLMs, why would we outsource it to an LLM that doesn't have our values encoded in it?