チャットGPT Chat GPT

 チャットGPT

チャットGPTを試しにやってみた。長生きの秘訣を聞いたら適度の運動、睡眠、食事、社会と繋がるなど、なるほどと思わせのる答がでた。しかし、２００７年発売の諏訪哲史著「アサッテの人」（芥川賞受賞）の批評を聞いたら飛んでもない答が返ってきた。

曰く、「アサッテの人」は三島由紀夫の作品で、芥川賞を受賞し、映画化されているという答であった。まったく見当違いの答だ。また、今私が執筆している時代小説の一部を「推敲して、もっと読者の心に響く文にせよ」と指示したら、これまた出鱈目の答が返ってきた。「である」調が「ですます」調に変換され、読むのに耐えられない駄文が返ってきた。

まだまだ、改良の余地があり過ぎる。知らない人が使うと危険だと思った。

Chat GPT

I have recently chatted with Chat GPT. To my question, “How can one live long?” it replied, “Moderate exercise, good sleep and diet, and social activities.” That was persuasive enough. 

However, to my question, “Give me a comment on ‘Asatte no Hito’ (an Akutagawa Novel Prize winning novel in 2007) written by Tetushi Suwa, a staggering answer was given. The AI said, “‘Asatte no Hito’ was a novel written by Yukio Mishima, won an Akutagawa Novel Prize, and was made into a movie. (It was not written by Yukio Mishima, nor made into a movie.) 

To another request of mine, “Rewrite the following passage of a novel so that it will move the reader,” AI’s answer was surprisingly foolish. All the “dearu” style of each sentence was changed into “desu-masu” style. (Generally speaking “dearu” style is for educated adults and “desu-masu” one for children.) The AI answer was intolerably bad and poor. It remains much to be improved. 

It can be dangerous too for those who are blind to Chat GPT’s faults.