Giant's Dream: AI Empowers the Future of Gaming

Giant's Dream: AI Empowers the Future of GamingAt the Yunqi Conference, Ding Chaofan, head of Giant Network's AI Lab, unveiled the technical details of their self-developed large language model, GiantGPT, and text to speech model, BaiLing-TTS, officially declaring the arrival of the "Games + AI" 2.0 era

Giant's Dream: AI Empowers the Future of Gaming

At the Yunqi Conference, Ding Chaofan, head of Giant Network's AI Lab, unveiled the technical details of their self-developed large language model, GiantGPT, and text to speech model, BaiLing-TTS, officially declaring the arrival of the "Games + AI" 2.0 era. During the conference, Giant Network showcased two self-developed large models: GiantGPT and BaiLing-TTS, along with their applications in the gaming industry.

GiantGPT: A Role-Playing Large Model That Can "Play Games"

As one of the first large models in the gaming industry to complete registration, GiantGPT has undergone continuous iterations and optimization since last year. Ding Chaofan introduced that GiantGPT possesses exceptional role-playing capabilities, vivid scenario reasoning abilities, and customizable long-term and short-term memory functions, enabling it to deeply support game scenarios. It can be considered a "game-playing" role-playing large model.

To train GiantGPT, Giant Network constructed a massive, high-quality dataset, integrating publicly available internet data with its own data, and establishing a complete and efficient data production pipeline. Currently, GiantGPT has been deployed in several of Giant Network's products, including "Zhuge." Companion-style AI NPCs built using GiantGPT exhibit distinct personalities, rich emotions, and adaptive long-term memory, providing players with intimate companionship experiences and improving user retention rates.

BaiLing-TTS: Breaking Down Language Barriers, Dialects Can Now "Speak"

Beyond large language models, Giant Network's AI Lab has launched the industry's first TTS large model supporting both Mandarin and various dialects - BaiLing-TTS. Leveraging the Chinese dialect system, Giant Network built a dataset encompassing 20 dialects, exceeding 200,000 hours of Mandarin and dialect data. They introduced several model-level technological innovations, enabling BaiLing-TTS to achieve zero-sample cloning of Mandarin and high-quality synthesis of dialect voices, including Beijing Opera singing.

Large Models Empowering Game Experience and Productivity Upgradation

During his speech, Ding Chaofan showcased Giant Network's explorations in large model applications, covering AI painting platforms, UGC script creation tools, personalized intelligent Q&A systems, and AI-native gameplays, demonstrating the transformative power of large models on both game experience and game productivity.

The implementation of large model technology in the "Among Us" project further exemplifies its potential to reshape game experiences and drive gameplay innovation. The game's UGC script creation tool integrates AI large model assisted writing and TTS functions, lowering the content creation threshold and sparking players' creative enthusiasm. The AI-native gameplay "AI Endgame Challenge" proved highly engaging for players, leading to a doubling of the game's related index on short video platforms and inspiring numerous players to spontaneously share interesting gameplays and strategies.

The core of the "AI Endgame Challenge" lies in Giant Network's self-developed Multi-Agent framework, encompassing both "collaboration" and "competition" characteristics. Constructing a balanced strategy based on a control system is crucial for forming high-quality collaborative and competitive paradigms. Moreover, deep player involvement is essential, requiring a focus on the flexibility and freedom in player task execution, as well as a robust operational mechanism to ensure a reasonable progression of the game.

The "Games + AI" 2.0 Era: Breaking Rules, Creating Limitless Possibilities

If productivity enhancement was the hallmark of the "Games + AI" 1.0 era, the gameplay revolution driven by AIGC technology propels the "Games + AI" 2.0 era. Ding Chaofan believes that "Games + AI" 2.0 will create a future game form: a world that breaks free from traditional rule constraints, where environments dynamically update based on player data feedback, storylines dynamically extend, and nonlinearity emerges through random triggers. This type of game would grant players immense freedom, even offering customized game content.

The "AI Endgame Challenge" gameplay demonstrates the feasibility of such a game form. Players can influence the environment through their actions, alter the course of the narrative, and enjoy high-quality interactive experiences throughout. This is not only a successful technological breakthrough and experiment but also marks the advancement of a new game form from concept to reality.

Comprehensive Deployment, Shaping the Future of Games

Currently, Giant Network has built a comprehensive foundation centered on self-developed large models, covering areas like large language models, visual content generation, speech generation, and AIAgents. Large models like GiantGPT and BaiLing-TTS have been deployed on a large scale in core game business scenarios, deeply integrated across game development, operations, distribution, and testing, forming an efficient production chain loop. Simultaneously, Giant Network is deepening the integration of large model capabilities into core gameplays, creating companion-style AI smart assistants, emotion-driven decision-making AI, and entirely new game paradigms based on multi-agent large models.

Looking ahead, Ding Chaofan emphasizes that Giant Network's AI Lab will be even more aggressive in exploring gameplay innovations driven by large models. He believes that the emergence of a native game world that engages players on a deeper level, naturally generating compelling and lasting content, is not far off.

As large model technology continues to evolve and gain wider adoption, Giant Network, with its forward-thinking vision and robust technical prowess, is spearheading the future of "Games + AI," creating immersive, intelligent, and infinitely possible game experiences for players.

Disclaimer: The content of this article is sourced from the internet. The copyright of the text, images, and other materials belongs to the original author. The platform reprints the materials for the purpose of conveying more information. The content of the article is for reference and learning only, and should not be used for commercial purposes. If it infringes on your legitimate rights and interests, please contact us promptly and we will handle it as soon as possible! We respect copyright and are committed to protecting it. Thank you for sharing.(Email:[email protected])