ChatGPTの実力を知っていただくには、実例を示すしかない。
たとえばJournal of Surgical Gerontologyから「査読で合格した論文をEditorialで取り上げたいからふさわしい研究者を推薦せよ」という以下のようなメールが届いたとする。
返事を二種類作らせてみたい。最初は「Editorialに推薦するほどの論文ではないよ」という返事。2つ目は「Editorialsitとして二人推薦する、その理由はかくかくしかじか」というものである。これが日本語の書きなぐりでできるのである。
☆☆☆☆☆☆☆☆☆☆☆☆
まず第一のやりとりはかくのごとし・・・・・
下のメールに返事を英語で作って
要点は1)editorialistを選ぶにふさわしい論文ではない、2)rate the priorityは2かな、3)tweetは不要
以上よろしく!
From: Journal of Surgical Gerontology
Sent: Tuesday, March 7, 2023 11:58 PM
To: Taro Saito, MD
Cc: admin@jsgeditorial.org
Subject: Your Editorial and Social Media Recommendations Sought – JSG 2022-12-3170.R1, Prognostic estimates of tumor-associated peroxisome in patients with adenocarcinoma of the lung
Title: Prognostic estimates of tumor-associated peroxisome in patients with adenocarcinoma of the lung Authors: Judge Smith
07-Mar-2023
Thank you for handling the above accepted paper. Please respond to this email by suggesting an editorialist for this manuscript if appropriate. It is the journal's goal to have all topics represented in its published commentary about its articles, including from your section.
If you think this paper is worthy of an editorial, please rate the priority of this paper on a scale of 1-5, with 5 being the highest priority. Any additional comments you have for Kelit Bushes about the worthiness of this paper as the subject of an editorial are appreciated in advance.
Also, please let us know if you recommend this paper to be tweeted on Twitter by the JSG Social Media Committee.
Your recommendations are appreciated in advance. Your editorial leadership of the journal is valued and appreciated. Thank you again.
Sincerely,
Geiten Verdenefine
JSG Editorial Office
コピペで貼り付けるだけ。すぐに帰ってくる返事がこれだ。
Subject: RE: Your Editorial and Social Media Recommendations Sought – JSG 2022-12-3170.R1, Prognostic estimates of tumor-associated peroxisome in patients with adenocarcinoma of the lung
Dear Geiten Verdenefine,
Thank you for reaching out regarding the manuscript "Prognostic estimates of tumor-associated peroxisome in patients with adenocarcinoma of the lung" by Judge Smith.
After careful consideration, I believe that this paper is not suitable for an editorialist, as it does not sufficiently contribute to the overall discourse in a manner that warrants additional commentary. Therefore, I would rate the priority of this paper at a 2 on a scale of 1-5.
Additionally, I do not recommend this paper for promotion on Twitter by the JSG Social Media Committee. I believe there are other manuscripts that may be more fitting for such exposure.
Thank you for your continued trust in my editorial judgment, and please let me know if you have any further questions or concerns.
Sincerely,
Taro Saito, MD
☆☆☆☆☆☆☆☆☆☆☆☆
さて、べた褒めしたいとき、推薦人を二人つけてみたいとき。そんなときは・・・
下のメールに返事を英語で作ってくれ。
要点は
1)editorialist候補の一人目は東京大学外科学教室のDr. Yukio Fujiki であり彼はperoxisome研究の第一人者だから。もうひとり挙げるとすれば、ハーバード大学外科のDr. Sachiyo Nomura であり彼女はcancer-stromal interaction の第一人者だから。
2)rate the priorityは5かな、
3)tweetで広く公知する必要あります。
以上よろしく!
書き込むとすぐに返事がくる。かくのごとし・・・
Subject: RE: Your Editorial and Social Media Recommendations Sought – JSG 2022-12-3170.R1, Prognostic estimates of tumor-associated peroxisome in patients with adenocarcinoma of the lung
Dear Geiten Verdenefine,
Thank you for reaching out regarding the manuscript "Prognostic estimates of tumor-associated peroxisome in patients with adenocarcinoma of the lung" by Judge Smith.
I believe this paper is indeed worthy of an editorial. I would like to suggest two potential editorialists for this manuscript. The first candidate is Dr. Fujiki Yukio from the Department of Surgery at the University of Tokyo. Dr. Fujiki is a leading expert in the field of peroxisome research and would provide valuable insights on the subject. The second candidate is Dr. Nomura Sachiyo from the Department of Surgery at Harvard University. She is a renowned expert in cancer-stromal interaction, which could offer a complementary perspective on the topic.
Considering the significance of the research presented in this manuscript, I would rate the priority of this paper as a 5 on a scale of 1-5.
Furthermore, I highly recommend this paper to be promoted on Twitter by the JSG Social Media Committee. It is crucial to share this valuable research with a broader audience, as it can potentially have a substantial impact on the field.
Thank you for your continued trust in my editorial judgment, and please let me know if you have any further questions or concerns.
Sincerely,
Taro Saito, MD
☆☆☆☆☆☆☆☆☆☆☆☆
返事としては十分すぎる内容ではないでしょうか、皆様。小生がChatGTPに与えた指示はぶっきらぼうだし、なにより日本語なのだ。それでも返事はこれだ。
僕はChatGTPは使えると思います。
☆☆☆☆☆☆☆☆☆☆☆☆
追伸(1):Journal of Surgical Gerontologyなんて雑誌はありません(笑)。お里が知れないようにでっち上げています。あしからず・・・
追伸(2):以上はChatGTP plus(GPT 4)を使って行った。GPT 3.5でも十分な返事が来るが、ややそっけない。この2つのバージョンの差をAI業界は驚異的に捉えているが、もともとのGPT 3.5でも「やるなあ!」というのが小生の評価であります。バージョンの違いを試されるのも面白いですよ。