I'd be wary of LLM-generated code, even if it seems to work in test cases. Are you sure it doesn't go wrong in edge cases, or have security problems? If you don't even know the language, you can't do any meaningful code review.