[CC#12] Generative AI and Video Games (Patreon)

While I was considering whether to use generative AI images in the playing card game I am currently developing, many people pointed out to me that they give a bad impression. I had heard such opinions about AI before, so I was not surprised. However, I also felt that some people might think I did not understand the situation well enough, so I would like to explain what I think about AI. On social media, differences in how words are taken out of context or nuances are interpreted can lead to unexpected arguments. For this reason, I generally avoid discussing such topics in depth there.

First, I would like to make one thing clear. I do not consider myself a creator. Even when someone calls me a game creator, it does not really feel right to me. I feel that toy maker is a more suitable description. I have felt this way for a long time. I have always been uncomfortable with calling games works of art, and the fact that I still talk about them as products comes from the same feeling.

Movies and novels contain the subjective perspective of their creators, which I think is then elevated into something artistic, into art itself. But what about video games? Is it the setting of the world? Is it the character design?

I am not denying game creators at all. I am simply talking about my own sense of discomfort. However, I believe this difference is a very important matter when it comes to how we approach game development.

I start by thinking about the gameplay itself. If possible, I have the intention of creating something different from what already exists. What are the elements that make players (users) feel that something is interesting? And does this new gameplay meet those expectations? Development begins with questions like these. At this stage, there is no world setting or character design. Only the feeling of playing the game is imagined in my mind.

When I feel that it might become interesting, I start adding details to it. I think about the design that is most suitable for this new gameplay. The story and characters are also created at this stage. As I have explained here, Mega Man was no exception. First, I built the game system, and the world of futuristic robots was the setting that best suited it.

That is why I do not understand where the authorial identity of a video game creator lies. Sometimes people tell me that they want to play a game made by AK. But I am at a loss because I do not know what kind of game they expect from me. Perhaps many of them mean a game like Mega Man. However, I made Cocoron as a game that was conscious of Mega Man in my own way, but it did not attract any attention. Its characters have pixel-art designs similar to those of Mega Man, and its gameplay includes character creation, an original idea of mine that had not previously existed in action games.

If I had to guess, I would say that people probably associate this with the world setting. However, I did not think of a world of robots because I wanted to make a game set in such a world. It was simply one idea among those that existed within my own experience and knowledge. Yet I feel that people imagine the world of Mega Man as the authorial identity expressed in my games.

The game I am currently developing is also an AKGame. However, it seems that a "playing card game" does not fit with my authorial identity. To me, Mega Man, Cocoron, KuruKuruPazuru, and SUIT4 are all games that were created in the same way.

In this sense, I think my approach to video game development differs from the way games are developed today. However, I do not think this is an old-fashioned way of thinking. That is because I already understood that my way of thinking was unconventional when I was making Mega Man.

My understanding is that images generated by AI are created by learning from works without the authors' permission, reproducing their characteristics and nuances, and using them to produce the desired image. However, I do not actually know in detail what happens internally. Still, I think the unauthorized use of works for training is the reason many people reject the use of AI images in games. Is that really the case?

Since I was very young, I have been exposed to Osamu Tezuka's works through television and manga. I unconsciously felt their dynamism, sense of speed, softness, and warmth, and I drew pictures by imitating Tezuka's lines and drawing style. There were also Shotaro Ishinomori, Sanpei Shirato, Mitsuteru Yokoyama, Hideo Azuma, Fujiko Fujio, and many others. There are so many works I have learned from. Of course, I did not obtain permission from any of these artists.

No matter how much people reject generative AI, it will probably spread throughout the world in the future. No matter how much they dislike it, it will come to exist in all kinds of places, including the things they love. This can no longer be stopped. And in ten years, young people of that time will probably accept AI images without any resistance.

"They can make pictures and animations so quickly, and they are incredibly beautiful, with realistic movement and nothing that feels unnatural. So why is it not acceptable?"

When that time comes, how should those of us who experienced this early period of AI answer this question?

Is unauthorized training really the reason people reject it? Could it be that people are actually reacting emotionally to something else?

My answer is that I believe the fundamental problem with generative AI images lies in the fact that we cannot tell whose work they are.

I think AI images have a very strong element of irresponsibility. They are works without an author. Even if they resemble someone's drawings or have the same story, the person who created them will not explain the reasons behind those choices. They simply rewrite the prompt and finish there.

Recently, something came back to me while looking at social media. When I was a child and had just started becoming interested in manga, I learned that manga artists had assistants who drew backgrounds and filled in solid black areas. At that time, I was shocked and disappointed, thinking, "What!? The artist didn't draw the backgrounds themselves!"

I feel that the resistance to AI is somewhat similar to this emotion. It made me very sad to learn that more than half of the drawings in the manga I was reading were not actually drawn by the artist themselves. However, once I understood that this was simply how manga was made, I no longer felt that way. Before long, assistants will probably also be replaced by AI. But nobody will reject it, right? Because that is how manga is made.

The influence of AI on video games will spread across a very wide range of areas. So, what kinds of areas will it enter? They include world settings, character designs, and background illustrations, as well as visual presentations that change each time by remembering the actions of the player (user).

Don't you notice something here?

The areas that AI will dominate from now on overlap with what game creators currently consider most important: what is commonly called "authorial identity." World settings, character profiles, and visual images will all come at us with an overwhelming impact. When that happens, game fans will all reject them at once. That is because one imitation after another will appear, seemingly copying the presentations and images created by game creators. And at a much higher level.

Images generated by AI are reconstructed from imitations and then output in a refined form. That is why they have no distinctive quirks. Ambiguous and unstable elements created by people are eliminated. You cannot feel the "human" element in them. The creator's quirks appear in games. There are many such quirks in Mega Man, too. But it is precisely because these ambiguous elements exist that the game never becomes boring, no matter how many times you play it.

However, there is something AI absolutely cannot imitate. It is the ability to think of new gameplay. Even if it reads every existing game and reconstructs them, nothing new will emerge. That is because ideas cannot be reconstructed. When the final form of something can be predicted, it is developed toward that expected outcome. But there is not even a definition of what new gameplay is supposed to be.

However, there is one correct answer when it comes to gameplay. It is that people find it interesting when they actually play it.

What is interesting? What kind of experience makes a person feel that something is interesting? It is impossible for current AI to imagine these things and come up with new gameplay based on them.

I have repeatedly said in recent years that while so-called big-budget games may look different visually, their gameplay all feels much the same. The idea is that making a game based on the gameplay of existing hit products leads to more stable profits than incorporating new ideas whose commercial success is uncertain. However, I feel that the situation will become even worse as the distinction between AI and human creators becomes increasingly difficult to see.

When I think about it this way, I wonder whether AI itself is not actually the "evil." Rather, the absence of a system for creating new gameplay may become the cause that prevents the game industry from developing further.

The playing card game (solitaire) I am currently developing includes one new gameplay idea. As far as I know, I have never seen it before. (If it already exists, then I simply do not know enough about it.)

I have been studying programming for several years, and the arrival of AI has helped me improve my skills to the point where I can develop a game in a very short time. AI is also useful for other tasks, such as processing large amounts of data and conducting random tests. In fact, the idea of creating rank badge images with AI came as an extension of this way of thinking.

If I were to use the term "authorial identity," then for me, it would be the "game system." And it includes new gameplay. That is the most important thing, and it is my original creation. Therefore, my thinking was that as long as that part was handmade, it should be fine to use AI wherever it could be used effectively.

Of course, the characters and their settings, and other elements that create the image of the game, need to use my own drawings. It was also important for me to draw the designs around the game screen in my own style, creating an image that evokes old gadgets rather than something sleek and sophisticated. The rank badges were supposed to be icons that players would feel happy to receive. I thought that shiny, sparkling badges would have more value as clear rewards than my clumsy drawings.

Introducing AI images was like having a manga artist's assistant replaced by AI. The perspective in the backgrounds was drawn accurately, producing a result that felt natural. It was also compatible with the artist's drawing style. That was the feeling I had.

I can fully understand the feeling of rejecting AI. Therefore, I decided not to use AI images, partly to avoid causing such feelings. I also believe this is the better choice for me personally.

To be honest, the speed at which AI is evolving is frightening. The other day, there was even an official statement about the need for initiatives to slow it down a little, but I doubt that will be possible. However, this is not necessarily bad news for me. That is because the game system, which is the most important part of my game development, is not being encroached upon by AI, at least for now.

(Automatic translation by AI)