The recent AI revolution, led by advanced applications like ChatGPT, is rapidly transforming a wide range of industries. Gaming and other virtual interactive experiences are among the first to be significantly impacted, and we can observe groundbreaking changes happening.
A recent tech demo called “Living Chang’an City” has been garnering attention. In this video, AI-powered NPCs can be seen roaming the streets of Chang’an City, each possessing unique identities and short-term and long-term goals. They engage in various life-like interactions, such as chatting, shopping and even falling in love.
The tech demo was released by Parametrix.ai, an AI unicorn company founded in Shenzhen, China in early 2019. Parametrix.ai has been dedicated to the fusion of gaming and AI and established deep collaborations with top-tier studios, making it a pioneer in the large-scale commercialization of AI within the interactive entertainment industry. The AI bots provided by Parametrix.ai have been successfully deployed in numerous leading gaming products, boasting tens of millions of daily active users. These AI-powered experiences have been well-received by players across more than 50 countries and regions.
According to Parametrix.ai, the recent AI NPC project is to explore the possibilities of combining their advanced AI bot technologies and Generative AI. The team developed GAEA system which brings Chang’an City to life as in the demo video. GAEA possesses the capability to create an AI NPC ecosystem, enabling virtual scenes like Chang’an City to operate autonomously.
In an interview, GAEA project leader Zhang Chi stated that GAEA integrates various model capabilities and has a longer development cycle than initially anticipated. The project is divided into two main phases: Phase one involves building a complete iterative scenario and technical framework, also known as GAEA 1.0; phase two focuses on enhancing GAEA’s story generation and natural language interaction capabilities.
During phase one, the team primarily relied on their proprietary AI BOT technology and long-term practical engineering capabilities, allowing the virtual scene of Chang’an City to operate autonomously. This stage not only requires a focus on enhancing the abilities of AI NPCs, including rich behavioral expressions and natural language communication to make them appear “more human-like,” but also on continuously improving the technical solutions and iterative roadmap, essentially bringing research concepts to reality. According to Zhang Chi, the latter is particularly challenging, as it not only involves considering the implementation of technology but also requires a product mindset that encompasses the entire NPC ecosystem from a systemic perspective.
Zhang Chi provided an example, explaining that making NPCs appear “more human-like” involves a variety of technologies. For instance, enabling NPCs to make reasonable actions based on common concepts like day and night requires the model to understand and reason about social common sense. To endow AI with long-term and short-term goals, the modeling of NPCs’ intrinsic motivations needs to be considered. Additionally, supporting a vast array of NPC behaviors calls for technologies such as multi-task reinforcement learning. Moreover, efficiently “remembering” and extracting information requires addressing the issue through unique encoding and decoding methods.
In phase two, with the system already operating seamlessly, the team introduced a large language model to enhance the system’s generative capabilities. As demonstrated in the original framework, the large model improved the existing GAEA system’s abilities in aspects such as goal decomposition, AI common-sense reasoning, and dialogue, further boosting the system’s performance.
Discussing next steps, Zhang Chi revealed that the team plans to continuously optimize GAEA’s technical details, utilizing large models to further enhance the NPCs’ ability to interact with real people and improve the story generation capabilities of the entire ecosystem.
The team is also currently contemplating how to deploy GAEA into lightweight, experiential products and establish collaborative opportunities.
According to the renowned Silicon Valley venture capital firm a16z, mastering generative AI technology is crucial for seizing the future gaming market. As game development barriers continue to decrease, the number of games released each year will grow, leading to the invention of entirely new game genres and the emergence of more adventurous and creative explorations.
Evidently, Parametrix.ai’s past experience in the “gaming + AI” area and the launch of GAEA have given them a strong first-mover advantage in this competitive landscape.
The release of the GAEA system may only be one of the earliest attempts in this wave. One thing is for certain: an exciting future gaming landscape awaits both game developers and players alike.
We know you don’t want to miss any news or research breakthroughs. Subscribe to our popular newsletter Synced Global AI Weekly to get weekly AI updates.