Will game companies impacted by AI experience "collective unemployment"?

Source: "DoNews" (ID: ilovedonews), author: Li Xinma

Image source: Generated by Unbounded AI tool

A year ago, the author once interviewed a senior practitioner in the game field. He said an interesting point of view: If you want to start a game project, you should first find a good original artist, because the game screen is not good-looking, the game will be cold first in half.

Unexpectedly, only one year later, the original painters became the group most affected by AI. Not long ago, Zhang Junbo, President and CEO of Unity China, said in an interview that many game companies have cut off half or even 80% of their art, and the next step to be impacted will be modelers. After that, game testing , programmers, and even game planning and scripting jobs, are likely to be affected as well.

For example, yesterday, at Baidu's Wenxin large-scale model technology exchange meeting, Baidu Smart Cloud used the newly released "Comate" code assistant to develop a "snake" game on the spot. Just input "canvas" and Chinese remarks, "Comate" can automatically recognize the semantics, generate the game canvas code, and then input simple words such as color, direction, and food, and then it can understand the instructions, complete the code, and generate the game.

In May of this year, Unity participated in GDC and released the "Game Industry Trend Report (2023)", which included AIGC as the future trend of game development for the first time. The report predicts that in the future, AI will further infiltrate into the game development workflow, accelerating every creative link such as images, audio, video, and code.

Image source: "Game Industry Trend Report (2023)"

Productivity determines the relationship of production. When AI can draw, code, or even do better than humans, the original production methods and organizational structures are destined to be changed. The original artist who was laid off was just the beginning. In the remaining game positions, the practitioners are also in danger, and they are full of worries about the same question-**Will I be the next one? **

To answer this question, we need a deeper understanding and thinking about the game industry.

Generally speaking, a game project group usually includes:

The project manager is responsible for the planning, execution and control of the entire project, and coordinates the work of various teams within the project.

Planning, responsible for game design, system design, task design, level design and scene design, providing clear game rules and gameplay for the game.

Art, responsible for the art production in the game, including the production of original paintings, models, textures, animations, etc.

Front-end, responsible for game front-end development, including page design, HTML/CSS/Java and other front-end code writing.

Backend, responsible for game backend development, including database design, API interface development, server-side code writing, etc.

Testing, responsible for the testing of the game, including functional testing, performance testing, security testing, etc.

Operations, responsible for the operation of the game, including user operations, event planning, data analysis, etc.

The original video games were created by developers line by line of code, but with the continuous accumulation of technology, various productivity-enhancing tools have emerged. For example, Photoshop for drawing, 3DMax for modeling, and the game engine represented by Unity, which has become the core tool for creating games. Two-dimensional pictures, three-dimensional models, audio, scripts, and animations, these assets are imported into the game engine after being completed externally, and even developers who do not understand programming can convert them into resources in the engine by clicking the mouse. Gradually integrate and iterate to the finished game.

From this perspective, the replacement of humans by machines has never stopped, but it has only come a little more violently. DoNews interviewed Yang Dong, Director of Unity China Technology Platform, to share the industry changes he observed.

"Original painting is indeed important, but currently some junior original painters can be replaced, just because of AI's ability to paint, which is currently the strongest category among all AIGCs." Yang Dong used to work in a large game company. The various processes of game production are very familiar.

"Junior original painters will first find a lot of references (references) according to the needs, and form their own ideas. They actually take a lot of pictures and put them together in various ways, and put them together into the thing that they finally imagined in their minds. , These cobbled together elements are also found out. From this perspective, the junior original artist is more like a material collector or reconstructor, rather than an original creator. The data in the AI training set must exceed the human brain and The pictures that can be stored on the hard drive can generate more beautiful or unique paintings, which will be more competitive. So they were replaced.”

In contrast, senior or master-level original painters have very strong original ability and personal style of painting. Although AI can also imitate the style of painting, it cannot produce original and unique thinking, so it is difficult to be replaced—but There are not many such people.

However, it will be a while before other gaming jobs, even entry-level ones, are replaced. "If it affects modeling, the quality of AIGC modeling is useless, it can only be used for reference, and it cannot be put into use immediately." Yang Dong said, "For programming, I think AI technology is important. It is not a replacement, but a positive assistance. For example, when I usually write code, I may have to write an algorithm from scratch or find someone else’s code to modify it. Now I can ask GPT directly, and it will write a version for you. I won’t use it directly, but I will first check if it is written correctly, and then try it out, most of the time it will make mistakes, but you can let it correct it.”

It is not difficult to find that the so-called substitution of labor is basically reflected in the emergence of better AIGC tools. The stronger the substitution, the greater the "unemployment risk". So in the core tool of the game engine, will there be a similar situation? Not long ago, Unity released a video called "Unity AI". In the video, a series of instructions are entered in the form of text boxes, such as "create a 3D female character" or "add two seconds of thunder", which makes people imagine , whether a sentence can generate the corresponding assets in the future.

However, Yang Dong said that the content in this video may not be realized until two or three years later: "The video expresses a vision, that is, a large language model can be used to help Unity developers automatically generate scenes, but now The big language model itself cannot directly generate a model. The generative model is another AI capability, and the ability is not very good. The big language model can understand what you are saying, and it can divide what you say into operational steps that Unity can perform , For example, how many models to put in the scene, where to put them, what kind of light to put on where, so that the bullets fired can collide with all objects, etc. It is not magical enough to say a word, just complete a game scene Now, it’s still too early to make a game out of one word.”

If AI reads enough text, it can learn to talk, if it sees enough pictures, it can learn to draw, if it plays enough games for AI, can it learn to create games?

"Theoretically, it is feasible, but the reality is rather skinny. We see that now feeding video to AI, AI automatically generates video, and it has not really been realized, and games are several times more complicated than text, pictures and videos. magnitude, so I don’t think AI can generate large-scale games in a short period of time.” Yang Dong said.

All in all, there will be shocks, and a complete subversion will have to wait.

But what is certain is that the efficiency of the game industry will become higher and higher, and the threshold for game development will also be lowered. In addition to worrying about how many people will be laid off, we must also see that AI has also raised the upper limit of game quality.

Not long ago, in his speech, Nvidia founder Huang Renxun showed a ramen shop with a cyberpunk feel in the game video, and the owner of the ramen shop inside seemed like a real person. In the video, Jin, the owner of the ramen shop, can chat with the player kai in real time, for example, to the player complaining about the crime near the ramen shop affecting his business, and answering the criminal leader Kumon Aoki when the player asks if he can help. , and guide players to the underground fighting gym to find the game Boss.

Image credit: Nvidia

According to reports, this is the "ACE for Games" service (ACE: Avatar Cloud Engine) launched by Nvidia for the development of character models in games. With the help of the large language model, the NPC can input necessary information such as background stories to the NPC in the game, and the NPC can interact with the player in real time, producing logical language dialogues, expressions, etc.

"This is what games will look like in the future," Huang said.

It is foreseeable that AI will be ubiquitous in the game industry, and the particularity of the game industry is that its core competitiveness should not be "cheaper", but more "fun". AIGC has lowered the threshold for a lot of professional work (such as original paintings) and raised the upper limit of content quality (such as NPC dialogue), which may stimulate more innovations in gameplay.

And innovation, or creativity, is one of the most difficult advantages of human beings to be replaced. Zhang Junbo said optimistically in the interview: "If AIGC really succeeds in overturning the process of game creation, it may bring another wave of startups and bring more choices and opportunities to the market."

In the relatively pessimistic forecast, that is, most of the human jobs and even creative jobs are replaced by AI, "but there must be many things that artificial intelligence cannot do, and there will still be many employees who will continue to do those things. No matter how AI is used, even if other jobs are replaced, there will still be programmers left in the world to solve the problems of AI programs.”