This research, using real-world examples, delves into how GPT-4, announced by OpenAI on March 14, 2023, interacts with Japanese translation and writing. Our findings are:
1) GPT-4 stands out as the top translation tool, with its Japanese translations scoring an average of over 80 points (A-).
2) GPT-4 sometimes “hallucinates,” meaning its output might not always be accurate or clear. Users should critically assess its reliability.
3) GPT-4 can generate text from scratch, making it challenging to discern if it’s human-written or AI-created. The old belief that “the writing mirrors the writer” doesn’t hold true since GPT-4’s introduction.
4) While GPT-4 can be a dedicated helper for Japanese learners, its grasp of grammar needs enhancement.
5) Be it for translation or other professional tasks, GPT-4 proves to be a reli- able “AI Copilot” and a handy AI assistant.
