Explain the AI strategy Unity CEO: combine reality with reality, and release AI tools this year to improve developer efficiency
Merging advertising platform ironSource, establishing Unity China, three rounds of layoffs, a total of more than one thousand people…… To expand profitability, Unity has introduced a number of reform measures in less than a year. From the financial report data, Unity’s measures can be said to be quite effective.
According to the 2023 Q1 financial report recently released by Unity, since the merger ironSource, Unity’s revenue has continued to strengthen. In the first quarter, the total operating income exceeded US $0.5 billion for the first time, up 56% year-on-year. The expansion of income also brought the Unity of continuous net loss after listing to level the loss,After two consecutive quarters of adjustment, EBITDA regained positive. Unity is currently seeking to further expand sales of digital twins and growth programs in China to further expand revenue. The excellent performance of the financial report was recognized by investors, which simultaneously boosted the sharp rise of the stock price.
As one of the leaders of the 3D content production platform, Unity has made a very positive performance on the current hottest topic of generative AI. As early as March, Unity released a short video of mysterious concepts, announcing that AI technology would be integrated into the engine. Unity CEO has previously said that it will launch a trading market for AI content and assets.
In this financial report, Unity specially opened the column “Gen-AI(AI era)” in the letter of shareholders, which further elaborated Unity’s AI development strategy. This new development attracted the curious attention of shareholders and analysts, and the AI topic became the absolute focus in the financial report presentation,It accounts for nearly 3/4 of the questions.
Game engine is a software hub in game development. If it can be combined with AI technology, it will undoubtedly have a profound impact on the way we develop games. However, it is still unknown what the final landing form of generative AI will look like without precedent in the industry.
What kind of play will the Unity of daring to eat crabs offer? Combined with the relevant contents of Unity’s financial report and Explanation Meeting, GameLook sorted out the AI blueprint in Unity’s conception.
What is the use of AI?
Unity said in its financial report preface that the emergence and rapid implementation of generative AI will be a “downwind” to boost the development of the industry, and the effect of this downwind will continue for several years. When asked what changes AI will bring to game development, Unity’s CEO John Riccitiello gives the following explanation: “First of all,Developers can use AI to create content faster, better, and cheaper. Secondly, people can use AI to achieve content that could not have been produced before.” In other words, AI can help game companies achieve “cost reduction and efficiency improvement”.
For what is “impossible content”, John Riccitiello gives a case like this: “For example, an NPC character has a conversation with you like a ChatGPT. The content of the conversation is not set by the script. They are very smart and ambitious. They have more free behaviors.In my opinion, this will bring about tremendous changes to the game experience and a large amount of growth.” In fact, after ChatGPT became popular, many companies have devoted themselves to the research of Intelligent NPC using generative technology.
Finally, Unity believes that AI’s natural language processing capability can lower the development threshold and help more non-professional developers and amateurs join the game development team.
“Combination of reality and reality”
Specific to the application form of AI technology, the core idea of Unity is “combination of reality and reality”. Unity believes that game content development will be combined with the use of two types of tools for a long time, namely, AI tools using natural language input and traditional deterministic (deterministic) editing tools.As an entry point, Unity will launch UI pages in natural languages this year to help users switch back and forth between more complex traditional editing tools and natural language AI tools to increase development efficiency.
As for why we didn’t choose to fully enter AI development, John Riccitiello to be frank, with the current AI capability, “one-click game generation” is completely impossible, and it is difficult to produce mature products that can meet the needs of customers for a long time. In Unity’s vision, in the current use scenario,Users will use AI as a quick draft generation assistant, and then transfer the results to a deterministic editing tool for further refinement.
John Riccitiello make an analogy with the work of journalists: “For example, you use ChatGPT to generate a press release, and then copy and paste it into a deterministic editing tool-that is, Word-for editing. Then you switch back and forth between Word and ChatGPT to modify the wording,Finally generate articles that can be officially published.”
At the same time, John Riccitiello also said that many artists or amateurs are not interested in using more professional code editing tools, and they prefer to use natural language to interact completely. The addition of natural language UI can also attract this part of users.
AI applications and challenges
In the financial report, Unity outlined several application fields of AI in the company’s products, including the Editor and Runtime of Unity engine, development analysis and distribution service Unity Network, etc. In addition, Unity plans to build the engine into a tool for developers to train AI models,Help developers “create, simulate and train data according to their own needs”.
In the long run, although Unity is extremely optimistic about the prospect of AI, John Riccitiello also bluntly say that there are challenges ahead. Among them, the biggest challenge is the complexity of 3D content development: “I have no doubt about the fact that natural languages can generate simple games, but the complex game content that we are used to consuming now will be difficult to train.”
From physical simulation to lighting effect, and then to the design of numerical system, the content involved in 3D game development is too complex. This kind of high-complexity content often does not have a unified solution, which also poses a high challenge to model training. However, Unity also expressed optimism: “This does not mean that we will not train such complex content,It means that large companies like Unity will have the strength to participate in the training process deeply, but companies with less strength can only stand aside.”
While Unity is exaggerating AI capabilities, other business engine competitors such as Unreal and Godot are still staying still for the time being, or are still watching market feedback to plan future strategies. On the Unity side, the faces of executives, analysts and shareholders participating in the seminar were filled with optimism, and the secondary market also cast confidence votes for Unity with the rising share price.
However, although it is good to remain optimistic, the final product experience is the golden standard for acceptance. As mentioned above, Unity’s AI natural language UI will be launched within the year, and GameLook will also track and report its use experience.
