[CEDEC 2023]Encounter AI and explore new expressions. Report of a session that looked at the history of spontaneous generation and talked about the problems and “possibility” of generative AI

Sedik 2023Today (August 24, 2023), the second day of“A Future Created With Alternate AI? Expanding Creativity Through Human-AI Co-creation”titled session

It has become a hot topic and a trendGenerative artificial intelligenceHowever, although it has all sorts of possibilities, there is also a growing concern about the impact that the manner in which human creations as a whole will be imitated and created on the work of artists, creators, and the artistic culture itself.
In this session, with topics such as current generative AI problems and possibilities and the relationship between AI and human creativity, since around 2000, we explored the application of AI to creative activities, tool production, and corporate R&D projects. CEO of Qosmo, who has been active as an artistMr. Nao TokuiKnowledge shared.

sessionProduction date by automatic generationAnd while looking back, talk about current problems and possibilities.
Expression and production using AI automatic generation has been a hot topic in the past year or so, but the same “procedural generation” mechanism has already been used in game development since the 1980s[انقر هنا], as described in the series written by Kaito Okutani. But even the history of self-generation was discussed already in the nineteenth century.

The mathematician Ada Lovelace, known as the world’s first computer programmer*, published a paper in 1843“If a computer can understand the rules of music, it can produce new music.”And automatic generation of music by programming. On the other side,Auto-generate does not create new expressionsHe said clearly. I thought I could imitate men so far as I had prepared, but I could do no more.

* Of course, when I say computer, I don’t mean the current personal computer, but the Analytical Engine, which is an early general-purpose computer.

Almost 100 years later, it was the mathematician Alan Turing who objected.If you can understand the rules of music, there is a possibility that new expressions can be born from thereThe “father of computer science” who invented the principles of computer, felt it from the stage of technology at that time.
Automatically generating music production would be realized in 1957, the year after the word artificial intelligence (AI) was coined. It is known as the world’s first piece of music composed by a computer.Eliac SuiteHe is.

Since then, in the 1970s, British artist Harold Cohen has createdBuild a system that mimics your drawing styleThen, in 2018, at Christie’s Worldwide AuctionsArtificial intelligence works to trade at high pricesIn some ways (and controversially), the art world has begun to talk about experimental methods and works that use spontaneous generation in various forms.

Based on the history of technical efforts through spontaneous generation, Mr. Tokoy created “DALL-E 2”, “Midjourney”, “Stable Diffusion”, etc.Differences between 2022 and beyondWe talked about
The art of artificial intelligence before 2021An attempt to explore new and unknown expressions by applying the automatic generation techniqueThe main one was also Harold Cohen“Illustrating the process of your artistic activity and the things associated with it”There was also an experimental use of this technique, that is, the precise and rigorous phrasing made possible by the machine, and the reconsideration of one’s style from a step back.

On the other hand, regarding the generative AI movement after 2022,Mimicry rather than sensitization – diminished as an attempt to imitate or reproduce the “similarity” of existing expressions (primarily illustrations)He said. Mr. Tokoy commented on these works: “Although I can create very precise pictures, when I look closely, they all look like this. I will.”

It is likely that the sense of crisis surrounding this is shared not only by creators, but also by many users who love their creations. There are quite a few artists and illustrators who are raising their voices on social media due to a sense of danger and avoidance.
On the other hand, the people around Mr. TokoyContemporary art and media artists see little dangerIt seems to show a reaction that is arguably the exact opposite. There are many possible causes, such as differences in field of activity and the way creativity is constructed, but this was taken up as an interesting phenomenon.
Also, for illustrators and illustrators who have concerns about AI art,“AI will not steal your job, on the contrary, the integration of AI may enable new expressions. Don’t worry too much, keep your head up and continue with your creative activities.”He said his thoughts.

At this hearing, Mr. Tokoy said:It is important to explore how current generative AI is encountered and used, not whether it is good or bad.He gave different examples. It was very interesting“Kakui Tanaka and Columbus in Creation”And“Unknown expressions generated from combinations of known expressions” (generated AI turns everyone into a DJ!)There are two.

“Tanaka Kakuei and Columbus in Creation”“exploration within a given expression”And“Expand the window and the same area”divided into two groups,What AI brings or does not bring to the OpenAI-like generative systemhad to explain.

What AI brings to the generative system similar to OpenAI,“exploration within a given expression”Corresponds to the active search and collection in the world of representation known through someone else’s past creations,“○○-like”High-speed, low-cost access is also possible when you want to express
Kakuei Tanaka, who aimed to build Japan’s high-speed network by building expressways and shinkansen lines.“On the transformation of the Japanese archipelago”And

What an OpenAI-like generative system AI cannot achieve (by itself)“Expand the window and the same area”. In other words, outside the learning area,AI is not very good at creating and expanding new types and modes of expression on its own.That’s what I’m talking about. This is because Columbus and many other navigators set out on the open seas in search of unknown continents.“Age of Discovery”expression is explained.

In “An Unknown Expression Born from a Collection of Known Expressions,” Mr. Tokoy, an artist who is also active as a DJ,The idea of ​​affirmative acceptance of generative AIIt was said.
Artists such as DJs and electronic music can reconstruct existing objects in the form of what is called sampling and remixing, such as making, mixing, and rearranging parts of existing songs, parts of songs, and instrument sounds. different song and create a new expression. In other words, it’s close to what generative AI does,Expression in generative AI (if it can be created rather than imitated) is DJ-likeI say so.

The “cutting and splicing” technique of such existing works to create new expressions is not limited to music, familiar to art lovers, subculture enthusiasts and artists, such as the “cutting” of . The responses of contemporary art artists and media artists that I mentioned earlier may be due to them having knowledge as the basis for their creativity.

At the end of the session, through the AI ​​DJ project, one of Mr. Tokoy’s activities, a method for exploring new expressions using the automatic generation of AI was presented.
Finally, as a summary, creativity is“The kind that looks inside the frame.”And“the kind that expands the frame itself”Generative AI has a high potential to accelerate the creativity of the first.It is important not to view the current trend of “OpenAI-like” generative AI (which imitates and fully reproduces human creations) as something that creates new expressions.With this, the session ended.