How AI Technology is Disrupting Game Development and Production?
-
Recently, at the China International AI Game Industry Conference, game industry professionals shared specific applications of AI technology in game development and production processes.
Xue Kena, Senior Product Manager of Microsoft Greater China Azure Division, mentioned at the conference that from the perspective of game development, "the entire game development toolchain has made significant breakthroughs based on AI." She added, "Combining OpenAI with translation, language, and other technologies can revolutionize the future of game development interactions. It may no longer require extensive coding; rich functionalities can be achieved through natural language."
Xue Kena cited the example of Minecraft applying AI technology in game development: "Through natural language interaction for construction functions, such as using natural language to communicate what to build, what environment to create, or what tasks to perform. Currently, AI character interactions and natural language functionalities have been implemented, making it a relatively mature and operational project."
Li Le, AI Product Lead from NetEase Fuxi, also shared how AI technology is being integrated into game development, using NetEase's Justice Online as an example. This game, dubbed an 'AI game,' incorporates AI technology in nearly all its core gameplay scenarios.
Li Le highlighted specific instances where NetEase applied AI in games, such as the 2017 implementation of a 'facial customization system' that allowed players to adjust facial features, makeup, and shapes. "Players only need to upload a photo, and AI automatically converts the 2D image into a 3D model, enabling players to complete facial customization in seconds, significantly improving convenience."
In 2018, NetEase introduced conversational AI into the game's parenting system.
Additionally, to maintain fair gameplay, NetEase applied AI in anti-cheat systems to identify various hacks. Combining AI with traditional anti-cheat methods enhances the ability to detect cheat variants. AI is also used in the game's matchmaking system.
In 2023, NetEase further utilized AI to create extensive game narratives and real-time cutscenes in its mobile games, pioneering 'text-based facial customization.' "Unlike image-based customization, this only requires text input," involving "models for text-image understanding, semantic comprehension, and image generation technologies."
Li Le noted that while the above examples involve integrating AI into games, Justice Online mobile can be considered an 'AI-driven game.'
In this game, "AI permeates nearly all player systems, from character creation—including NPC appearance and dialogue—to game narratives, combat, and even social and exploration gameplay. We embedded over a dozen AI capabilities into Justice Online, deeply integrating AI into the game. From the moment players create their characters, AI accompanies them, offering a novel experience throughout their journey."
For instance, in narrative generation, "while older games might have static backgrounds or limited variations, AI now enables pre-generation of extensive cutscenes. For real-time scenarios, AI synthesizes expressions, movements, and lip-syncing to enhance player immersion."
This AI-driven approach to narrative generation is "low-cost and fast. We’ve already generated 100 hours of expression animations for Justice Online mobile, significantly reducing development costs."
The game also employs AIGC (AI-generated content) technology. For example, "players can use AI to compose Song Dynasty-style poetry while enjoying scenic views, with 40 poetic templates available. The generated poetry can also be recited using cloud-based synthesis technology."
AI has also penetrated game art design. At the conference, Chen Hong, an original art designer from Boke City, shared the pros and cons of AI-generated art.
Chen Hong noted that before using AI, designing a set of 7-8 game characters required a team of four, outsourcing, and refinements, taking a full month. With AI, "line art can be extracted in minutes, and tools like Stable Diffusion, ControlNet, and Photoshop reduce the process to about 15 minutes, compared to a day traditionally."
"We integrated Stable Diffusion to streamline refinement, allowing two in-house staff to handle the process. Later, with MidJourney’s Niji V5 model, I alone managed character design for the project, still within a month."
Beyond characters and scenes, game projects often require card designs. "Post-launch, we needed ~150 card images every three months. Traditional methods were time-consuming, so we used MidJourney for generation and Photoshop for adjustments. A small tool automated overnight generation, yielding 500–600 images, from which we selected and refined the best."
"AI excels at mass production, delivering faster and better results."
Chen Hong added that AI art is useful not only for large projects but also indie games. "Using AI for early demos is highly efficient."
However, he cautioned that "while tools like MidJourney provide references for色调 and completeness, they offer limited help for highly specific early-stage designs. Some complex requirements remain unachievable by AI, and manual refinement is often needed for细节结构."
Lastly, Ren Xianmin, Partner at Jiuyu Education, proposed a cross-disciplinary perspective on future AI applications in games. He suggested that "as AI advances, games could offer personalized AI companions tailored to each player." By extracting data on players'内在状态, developers could customize content, particularly in RPGs, card games, or strategy games with strong社交属性 and UGC elements.