It is said that at this year's ChinaJoy Conference, China's "Game Spring Festival", the audience first participated in a small survey before entering the venue: choose yes or no on the question of "will AI destroy humanity".
I have to say that game fans are really familiar with this problem, after all, among the many apocalyptic games they have played, machines or robots representing AI have replaced humans as the masters of the earth.
Even if it jumps out of the game, AI replaces human labor, allowing large companies to reduce costs and increase efficiency much faster. Returning to the game industry itself, cloud vendors, AI large model companies, and even infrastructure companies such as game engines are also integrating AI and large model technology into their products to make game production easier.
AI seems to have the potential to usher in a new revolution for hundreds of billions of dollar game products?
Prior to ChinaJoy's opening, game engine giant Unity hosted its 2024 Worldwide Developers Conference in Shanghai. At the conference, Unity China CEO Junbo Zhang was interviewed by Geek Park about the impact of AI on the gaming industry and Unity's goal of "spillover" in more physical industries.
Zhang Junbo believes that if AI technology can indeed produce the current level of games, large game companies may indeed face great changes, game teams will be more miniaturized, and the threshold for truly good ideas will be lowered. However, if the revenue scale of the game market does not grow with it, the game industry will face a "new balance".
Outside of gaming, Unity's aggressive efforts in automotive intelligent cockpits and industrial digital twins have given people a glimpse of the game engine company's new growth curve and potential.
The following is the transcript of the live interview, edited by Geek Park.
Junbo Zhang, CEO of Unity China, delivered a keynote speech at the Unity Worldwide Developers Conference
AI may steal the job of code farmers, but it's a good thing for game creators
Geek Park: Now because even though Unity has launched Sentis, and Google Cloud has said that it has AI to help game developers, it can't completely replace a lot of productivity to save time for game teams, do you think in 3-5 years, is it possible that AI will have some impact on the production organization of games?
For example, through AI, what a team of 1,000 people used to do in a game company can be done by only 30 people in the future. Will this be the case in the future when giant game teams become countless small studios?
Zhang Junbo: Yes. If AI capabilities become and produce the same type and quality of games, if fewer people are needed, there are two problems, there will be fewer practitioners, and the other question is whether we can produce more games, and then everyone can choose more, and not everyone can play one game, there are more good games, and everyone's creativity will be released.
In the past, you had to make a game, and this game needed 1,000 people, and to set up a project, it cost 2 billion yuan, and there were very few people who could get these 2 billion, but if it became 20 million now, it could come out and it could be launched. Therefore, AI capabilities can make everyone's creativity more possible and achieve at a lower cost, and players can play more. AI will create more creators, and before it was the ratio of creators to programmers, and maybe the ratio of artists to art was smaller, because you need a lot of people to make an effective idea, but now there are creative people who can use their skills more.
You originally wanted to make a company with 5 planners and 30 to 50 programmers for a game, and you could only listen to one person in these 5 plans, and that person had the final say. In the future, I will drive one myself and do what I want.
Now the game can do this because 1000 people only have a budget of 2 billion. That's why this effect is made. But whether this effect is the termination effect of the game, it is possible to say that I will still invest 2 billion, and it will be even more different if I make it, I am a simulated earth, and I may earn more. In general, this is a cost-effective price, that is, you put in, if AI improves production efficiency, it is improving the output per unit of input, and everyone can play more good games.
In the entire game market, everyone is willing to spend money, and many gamers can only charge 68 yuan a month. So no matter how many games you play, I only have one 68. The whole market is here, and there will be another equilibrium in the end. Before 68 can only be the top 10 games, everyone controls 80% of 68, will there be 1000 games in the future, it is very evenly distributed here, 80% is eaten, everyone has a better chance to earn more, AI can improve production efficiency, this is a good thing for the game industry, programmers' jobs may be robbed, but creators' jobs will not be robbed.
Geek Park: CoPilot on Github can help people write code, which is equivalent to a person who can generate a small game, and for WeChat mini games, its code ability is almost OK, which is equivalent to a person, as long as his creativity is enough, it is easy to make a small game like a sheep.
Zhang Junbo: Yes. Faster cycles. Ideas used to be too expensive to realize, but now they are cheaper to realize.
Wherever there is a screen, we will go
Geek Park: Today, we also revealed some information about the non-game part, one of the biggest highlights is in the automotive industry, where Unity accounts for 85% of the intelligent cockpit. But why is Unity investing so much in an industry with a long investment cycle?
Zhang Junbo: For the automotive industry, we have a long-term view. Because Unity is ultimately a screen, and where there's a screen, we go to render things. Nowadays, car companies are full of screens, and these screens want to render content, and 3D interactive content is a very natural. So this is where we look at the automotive industry, and this is the future.
It's just a start. So the application of intelligent cockpit, such as car control, ADAS, and map is basically a must-install thing in the car, and the map is still a little more advanced, but ADAS is very basic, using 3D to make you feel better and look better, this is just the beginning, we pass these we hand 3D technology to the car, those screens in the car, and eventually everyone will be liberated from driving, he has more time to consume more content, it can be a game, it can also be a variety of experiences, the screen in the car is large, the sound is good, the experience is good, It feels much better than when you play a mobile game.
So this is where I think in the future, HMI (human-computer interaction) is the first step, and the OEMs are responsible for it. So we have to talk to the OEMs, this is a different business model from the traditional game industry, and there is another way of cooperation with the OEMs, our first step is to go first, and we have so far in many cases, because the industry has risen up on its own, and then we sometimes have some OEMs that we take the initiative to go to PD, to persuade them, to do POC for them, to persuade them to do it.
But more companies know that he is going to go to 3D, and the ecology on the car, Android, 8155, 8295 (chips) are closer to the mobile terminal. So if it's going to do 3D content, he will default to it, or prefer Unity, because Unity is easy to use and powerful first, and the technicians in their company, or two people from the game industry outside, can start.
The traditional industrial software for making cars can only be used by a few people, and only a few people can use it for vehicles, and the cost will increase. With Unity, the cost of trial and error is low, and he can use it right away. So what we see, why we have a high acceptance rate and high coverage, a lot of reasons are because of the value of the engine itself, its ecology.
Unity 在智能座舱市场占有率到到 85% ｜Unity
At present, the scale is not large, but if you put it out globally, there are 70 million to 80 million cars a year in the world, and car companies have their own models. So if we can achieve such a share as mobile games, it is actually still very large, because now not only electric vehicles will go to the intelligent cockpit, even some oil cars, and the Mercedes-Benz E series that Daimler is talking about today is a fuel car. Of course, you can't expect the petrol car to play games on it, because the battery will be exhausted for a while before the tram can play games, and the smart cockpit still uses 3D technology.
So the plate is quite big. Moreover, the apps and mobile apps that are derived from this intelligent cockpit to the on-board car, as well as various high-value products covered by the future car will also be covered. So then there is a possibility that there will be a car configurator, which is derived from this series, as well as the car user manual. So we think the car is a very worthwhile thing to do, and it doesn't necessarily have to be smaller than a mobile phone.
The stock of cars, although only 80 million per year, but for a car in ten years, China has hundreds of millions of dollars in stock a year, and there are more and more cars. So this market is no smaller than mobile phones.
Geek Park: It's not 3-5 years when Unity thinks about cars, it's longer.
Zhang Junbo: There will be more screens in the car than at home and on mobile phones combined.
Geek Park: Unity also mentioned some digital twin projects in industry and architecture, what is Unity trying to achieve? Or maybe it was because the BIM industrial software was too complicated before?
Zhang Junbo: It's not necessarily that it's difficult to use, but its use is not quite the same, now to do digital twins, many times everyone's goal is not to use it, but to use it for operation and maintenance, to do operation and maintenance, a factory, or a building, the equipment and CAD data in it come from a variety of things, because it has to implement a logic on it, a production line, a plant, the structure is made according to BIM, and the pipeline may be made by another software. The robot and the production line and the assembly line are made by another provider, and you have to piece together the machine to make a delivery.
From a machine B to take it from here and there, and then to go up there, it's probably done by two different things. The software it conducts is CAD software, and it will have its own simulation capabilities, but it is the ability to accurately simulate its individual equipment, but if you want to say I put this bottle from here to there, can the robotic arm there pick it up, you really have to simulate, although the logic is not complicated, but you have to string it together.
So the good thing about Unity is that I string things together, from different industrial software and CAD software, I put it in a visual place to string it together, and I can put this screen down, it's actually a rope here, there's a sensor, I can take the content of this sensor, and then it becomes that I want to program, I have a business logic, I have a state transition, I want to program. Unity has a script that is great for doing these business logic transformations. So that's what is used to do digital twins.
This is what we see later. Of course, everyone has gotten rid of just looking. How to operate and maintain the factory? In the future, some digital twins may have a central control room, and there may be many central control rooms, you see, but the thing that the central control room looks at is a bird's-eye view, a bird's-eye view of the state of the entire factory, if it is all green, there is nothing to do, because everyone monitors to see where something went wrong and called the police, and if I have a bird's-eye view If the alarm is called, my 3D product will jump out very intuitively. The traditional one says the fifth way of the seventh workshop, where is that, but if you have a bird's-eye view, you will know it at once.
Once you know that, you have an inspector, or a robot, and show me what's going on. The robot is wearing a glasses, and when he goes over there, the local information he sees with the Pad is linked to the central control. So what we see is this 3D twin system, which is used to allow everybody to see things through a unified system.
If it is patrolling in the city, the screen that the police see in the police car should be the same as the state that it looks at in the headquarters, but each person has a different perspective, and then everyone can collaborate through 3D, which is our vision of the future industry may be industry or architecture, and its twin can have this application scenario.
Geek Park: Do you think the Unity Engine has lived up to our expectations in the non-game business?
Zhang Junbo: We have achieved some results, but we always feel that we are slower than expected, many things are thought of beautifully, but there are various challenges in implementation, we believe that we are still making steady progress in the right direction, especially in the period of unity engine, we have also experienced the bubble of the metaverse from 2022 to the present, when there is a metaverse, everyone says a lot, but in the end many things have not landed.
The current application scenarios we see, 3D, have not been reduced, but just in the past, from a swarm of people, everyone was eager to grab the city of the metaverse, built a lot, everything had to rely on the metaverse, and created something very virtual and did not happen.
Under the previous assumptions, we may have set some unrealistic goals, but now that we look at it, we can actually achieve your ability to reduce costs and increase efficiency, and can help you generate and operate and maintain, and everyone is steadily investing there, which is also very consistent with the R&D route and technical route of our unity engine.