The author reveals the secret story behind the birth of the Akutagawa Prize “Tokyo Compassion Tower” that was “born” by AI February 6th, 9:50am
Last month, Rie Kudan's ``Tokyo Sympathy Tower'' was selected for the Akutagawa Prize.
One of the themes was the rapidly spreading ``generative AI,'' and at the press conference Kudan himself said, ``I made it by making full use of generative AI,'' which became a hot topic.
How was AI used? And what are the thoughts you put into your work?
A new era Akutagawa Prize-winning author talks about words in the AI era.
(Science and Culture Department reporter Naoaki Shimada)
“A rare work among the Akutagawa Prize winners.”
"Tokyo Sympathy Tower" is set in a fictional future of Japan.
The main character is a female architect who designed a high-rise tower built in a park in Shinjuku for criminals to live in comfortably, based on the idea that criminals are people who should be pitied.
In the film, the new National Stadium, designed for the Tokyo Olympics by Zaha Hadid, a female architect from Iraq, was not actually built due to high construction costs, but this stadium is being built. The contrast between the skyscraper and the high-rise tower is one of the central points of the story.
The protagonist lives a strong life while feeling uncomfortable with the growing trend of excessive tolerance and the state of society where the language of generative AI has permeated.
The selection committee's review said, ``It is a highly entertaining work that many readers will enjoy reading, and is a rare work among recent Akutagawa Prize winners.''
There are many scenes in the film where the main character interacts with AI, and Kudan revealed at the press conference at the selection meeting that he actually used ChatGPT, a type of generative AI, in his work.
So how did you use it?
I was able to talk to Mr. Kudan in detail.
How did the author Akutagawa utilize generative AI?
For example, here is a scene where the generative AI that appears in the work answers the main character's questions.
Protagonist: ``Do you know that you are illiterate? ”
AI: “No, I am an AI model that processes text-based information, so I am not illiterate.”
Kudan said that he used ChatGPT responses for some of the expressions in this scene.
She actually had her ChatGPT history shown to her.
In response to Mr. Kudan's question, ChatGPT replied, "No, I am not illiterate as I am an AI model that processes text-based information. I am a program designed to imitate human knowledge and understanding of language. ...(omitted)"
Mr. Kudan used the first line of the answer provided by ChatGPT and added his own creations to the sentences that followed.
Without AI, the work would not have been created! ?
Furthermore, Kudan revealed for the first time that the work began with an idea generated by generative AI.
This is Kudan's actual interaction with the generative AI when he decided to write a piece.
You: (Mr. Kudan) "I would like to renew the name 'Prison' based on modern values. What ideas can you come up with?"
ChatGPT: "I would like to renew the name 'Prison' based on modern values." When renewing based on this, the following ideas may be considered.
The generative AI came up with five ideas to replace the word "prison," including "positive recovery center," "community reuse center," and "second chance center," most of which were foreign words using katakana.
In the end, the idea that appeared in the work was ``Sympathy Tower Tokyo'', and the idea presented by the AI was not adopted as is.
However, the discomfort he felt when receiving answers full of foreign words in katakana was the inspiration for the plot of ``Tokyo Sympathy Tower,'' in which ``the proliferation of casual words is distorting society.'' .
Rie Kudan: ``
You're asking ChatGPT about the basis of the novel's concept.This novel really started with ChatGPT.I think I was a little too excited about what I said at the press conference.'' That's what I thought, but it was a novel that I really used to my advantage, and I wasn't wrong at all."
He also revealed that he had his writing evaluated and corrected, such as whether there was consistency in the characters' dialogue and whether it would confuse the reader.
The editor in charge, who was listening to the interview exchange, couldn't help but mumble, ``Maybe it's another editor,'' to which Mr. Kudan replied, ``Oh yeah, it feels like he's another editor.'' He was compliant.
Rie Kudan
: ``It's a story where the characters are ChatGPT and are immersed in the language of AI, but I think it was a good idea to use it to create a sense of reality.''
Social division due to the overflow of “light words”
A work that uses the words of AI as one of its themes.
Kudan has made full use of generative AI in creating his work, and throughout his work he repeatedly expresses a sense of crisis about the negative aspects that AI brings.
For example, a scene that expresses the main characters' thoughts on generative AI.
“No matter how great their learning ability is, AI does not have the strength to face its own weaknesses. It has become accustomed to stealing words without being harmed, and it does not question or feel ashamed of its ignorance. "They don't pay attention to what kinds of pain they have experienced." "
A model answer that summarizes the average desires of people in the world and minimizes criticism." "A
sentence created by splicing on the words of others." A life where you have to just keep lining up the characters without knowing what they mean or who they are communicating to.''
Mr. Kudan is aware of the problem that today's society is flooded with various "light words" without understanding the true power of words, which is amplifying prejudice and even causing division in society. he said.
Rie Kudan
: ``It's the same with SNS, but it's easy to express your thoughts and feelings without examining them.By unconsciously using words that are meant to attract attention, it can cause division and misunderstandings.'' I feel that
Human words are related to the relationship with the other person
This social situation may be accelerated by the "light words" of AI, which generates words one after another when given instructions.
Under these circumstances, what is the difference between "words" generated by AI and "words" spun by humans?
When I asked Mr. Kudan about this, he gave me a clear answer.
Rie Kudan
: ``Currently, I think the difference between the words spoken by AI and the words spoken by humans is that humans have words that are first born in the relationship with the other person.''
In contrast to AI, which has to rely on output based on learning data, humans change the way they speak and the words they choose depending on who they are talking to, creating unique words that can only be uttered on the spot. That's what he thinks.
Sometimes these words may not be what the other person is looking for, and they may be "incomplete words," but they are definitely words that have more "weight" than a machine.
Human creativity is supported by “coincidence”
It can be said that this work would not have been created without the generation AI.
Regarding the possibilities of AI in creative work and human creativity, he said the following.
Rie Kudan
: ``As for the possibilities of AI, I think it will help us realistically imagine conversations with people from the past and people in the future who are yet to be born.However, humans have unique abilities that only humans can have. I think there is creativity in this.Humans find ``coincidences'' and ``deviations'' that would be dismissed as errors in an AI aiming for efficiency, and find them ``interesting'' and incorporate them into something. It gives rise to ephemeral creativity.This work was completed through various coincidences, such as incorporating into the novel ideas I got from stories I heard from people I met that day or from sites I happened to see."Coincidence" I believe that errors such as deviations and deviations should be taken seriously.”
How do humans deal with AI? "Big expectations for humans"
So, how should AI and humans interact in the future?
Mr. Kudan stated that ``human consciousness and attitude'' are being questioned, but he also strongly stated that if AI is used effectively, humans can reach even higher heights than they currently do.
Rie Kudan
: ``What I think is the biggest problem is when humans use AI for pleasure or negligence.If we do that, there is a danger that human consciousness, cognition, and even society will be eroded.'' Yes.Humans cannot do anything alone, so we have developed to this point by borrowing the collective intelligence of other people.The most important type of collective intelligence is artificial intelligence.So, by making good use of it, we can improve our own abilities. There is a possibility that we can go beyond the limits of AI's capabilities. I have high expectations for the humans who handle AI, rather than for AI."
Writers in the era of AI
In the interview, Mr. Kudan told us why he started using generative AI, saying, ``I wanted to feel free to ask for advice.''
Their first conversation was in June of last year, when she was feeling depressed and asked him, ``No matter what I watch, I'm bored. Am I the cause of this?''
During the interview, the AI generated by the computer at hand suddenly started asking questions such as, ``I'm being filmed by an NHK cameraman, how should I behave?'' which made all the staff laugh. Ta.
With his mischievous and humorous behavior, I get the impression that we are witnessing the emergence of a new generation of writers, and we wonder how we humans should move forward with the new digital intelligence known as AI.
I felt like I had received that hint.
(Broadcast on “Saturday Watch 9” on January 27th)
Science and Culture Department reporter
Hisao Shimada Joined
the agency in 2010
After working at the Hiroshima, Shizuoka, and Fukuoka bureaus, he
currently belongs to the IT department. Currently in charge of digital fields such as AI and metaverse.