Giant Network showcases the first batch of innovations integrating gaming and artificial intelligence at the Yunqi Conference

According to news on November 2, Giant Network’s first batch of “game AI” results were unveiled at the Hangzhou Yunqi Conference. Ding Chaofan, head of the company’s AI laboratory, shared the results of the project in the forum. The overall research and development efficiency of the game has been greatly improved. promote. The company's "game AI" goal is to revolutionize game production and promote innovation in gameplay.

AIGC has carried out production innovations and greatly improved R&D efficiency

Giant AI Lab has established a set of "game AI" efficiency enhancement platform based on the core production elements of the game. The platform covers images, video, audio, code, text, models, strategies and other fields involved in game development, operation, distribution and other aspects, and uses different AI technologies to build an industrialized production pipeline

Take iMagine, the AI ​​painting platform that Giant cooperates with Alibaba Cloud artificial intelligence platform PAI, as an example. The platform can provide AI solutions for the company's art team to achieve industrialized production while reducing local complex processes. Users only need to open a web page to use it

According to Ding Chaofan, the efficiency benefits of AI painting within Giant are very considerable. Overall, we have achieved 50%-70% savings in character and scene original paintings, and more than 80% efficiency optimization in UI and icons. In some projects, the internal UI and icons have even been completely handed over to AI for rendering.

AI painting not only improves efficiency, but also plays a huge role in improving the quality of art. Take the mobile game "Space Killer" as an example. After careful processing by AI, the three-dimensional sense of the original character painting has been greatly enhanced, solving the problem of the previous hand-drawn being unable to achieve standard quality.

Another value of AI painting is the subversion of traditional workflow. It changes the way of communication and collaboration between planning and art, and greatly reduces the time cost of communication and collaboration between key R&D positions.

In addition to AI painting, Giant AI Lab has also conducted research and development on AIGC video, voice and code, and has achieved some results. In terms of AIGC video, the company has developed its own video style transfer tool, which can be widely used in video promotions to provide players with materials, thereby producing a fission effect; in terms of AI voice, the company has launched the Bailing AI voice generation platform , through AI voice generation technology, the giant player group can hear more realistic NPC voices

In terms of large models, the laboratory has also done a lot of application exploration. For example, the company launched a large model of the CodeBrain code layer within the company, which combines the company's own data and self-developed compiler plug-ins to provide efficient code generation assistants for the company's internal developers, increasing the research and development efficiency in this area by up to 30%. -40%.

Break content homogeneity and promote gameplay innovation

Regarding the value of AI, there is a consensus within Giant Network that reducing costs and increasing efficiency is only a means, and promoting innovation in game play is the ultimate goal. Giant AI Lab also regards "breaking content homogeneity and promoting gameplay innovation" as its long-term goal, and currently has phased results in multiple game projects.

Take the new game "Masterless Throne" independently developed by Giant Network as an example. This Western-style magical-themed camp versus MMORPG mobile game is expected to be launched in the second half of next year. Based on the traditional camp versus MMO game, the AI ​​Lab developed a large model-driven AI NPC ecosystem

Ding Chaofan showed a real-machine demonstration of the ecosystem on the forum. He said: "In "Masterless Throne", we can see that the large model has the ability of long-term memory, planning and decision-making, and self-reflection. Each AI Agent controlled by the large model and capable of interacting with the environment builds an An intelligent agent ecosystem that never stops running. Each AI NPC in it can provide players with the most realistic and delicate interactive immersive experience. This intelligent agent ecosystem will add more randomness and imagination to the virtual society."

In another relaxing and casual mobile game "Space Killing", the development team used AIGC technology to achieve innovation in gameplay. "Space Killing" is a casual-themed, multi-player social reasoning mobile game independently developed by Giant Network. Based on the ability of large-scale models, the team creatively created an AI reasoning theater as the built-in gameplay of "Space Killing". Players need to engage in human-machine dialogue with the AI ​​in a given case context. The AI ​​presets the backgrounds of different characters. Players need to judge the real murderer based on the AI's answers and the case process until the case is solved successfully

"As a player, you found a defense point of AI based on your own reasoning, and finally you succeeded in breaking through the level. In fact, it is an AI Native Game (AI native game), that is to say, it is an AI native game. Driven gameplay. If the large model ability in this gameplay is removed, the entire gameplay will collapse and this gameplay will not be established. Therefore, the key to this gameplay is the large model, and the intervention of AI makes the player's imagination richer. "Ding Chaofan introduced.

It is worth noting that the company’s oldest and flagship Zhengtu IP track is also actively embracing AI changes. The team has used AI to build a risk control system for the entire Zhengtu IP track. This system provides comprehensive monitoring and defense for the game ecosystem. It has achieved initial results in tracking abnormal transactions in gold farming studios and protecting the economic and ecological balance of the Zhengtu series of products.

In Giant’s view, the industry is ushering in the era of “Game AI” 2.0. The "Game AI" 1.0 era focuses on improving productivity, while the "Game AI" 2.0 era relies on AIGC's technology to completely liberate productivity, release creativity, and promote gameplay innovation. This will strengthen the core competition at the bottom for independent research and development game companies. force.

The above is the detailed content of Giant Network showcases the first batch of innovations integrating gaming and artificial intelligence at the Yunqi Conference. For more information, please follow other related articles on the PHP Chinese website!