[AI Revolution](18) For More Fun Games, 300 People Deployed for AI Research... NC

NCSoft Established AI Research Organization in 2011
Developing Various Technologies Including AI Translation and Digital Humans
Continued Consideration of AI Ethics

The appearance of Kim Taek-jin, CEO of NCSoft, who appeared as a digital human at GDC 2023.

Last month, Kim Taek-jin, CEO of NCSoft, appeared at the world’s largest game developer conference, GDC 2023, held in San Francisco, USA. However, the Kim CEO on the screen was not the real person. It was a digital human perfectly reproducing his face, voice, and manner of speaking. Developers attending the event were amazed. It was not merely a physical replica but another version of Kim CEO capable of real-time communication.

12 Years of AI Research and Development

NCSoft’s passion for artificial intelligence (AI) development is already well known. In 2011, NCSoft established the first AI research organization in the gaming industry. From a team of 150 members in 2019, it has doubled in size to 300 members this year. Recently, the AI Center, NLP (Natural Language Processing) Center, and Applied AI Lab were integrated into a unified 'NC AI R&D.' The research fields are diverse, including video, voice, graphic AI, and the development of large language models like ChatGPT.

The sole reason NCSoft is dedicated to AI research is to create more enjoyable games. The digital human technology showcased at GDC 2023 is one of the technologies aimed at making games more fun. NCSoft’s digital humans do not leave users alone. They listen, speak, move like humans, and decide their own actions. Games offer a higher level of enjoyment when played with others rather than alone.

Currently, NCSoft is fully committed to developing digital humans that can naturally converse and interact with people on various uncertain situations and topics. The research team possesses all the technical elements necessary for digital human development, including AI and NLP. At GDC 2023, NCSoft demonstrated Kim CEO’s digital human directly acting within the upcoming game ‘Project M.’

Integrating Global Users with AI Technology

NCSoft is applying various AI technologies to its games. A representative example is real-time AI translation.

Lineage W is well received in 12 countries worldwide. Due to the MMORPG (Massively Multiplayer Online Role-Playing Game) nature, frequent communication between players and strong relationships such as guilds are necessary, supported by AI translation technology. It even translates Lineage-specific terms that other translation technologies cannot recognize. This fosters bonds among users across borders.

NCSoft has also introduced events applying AI reinforcement learning models to Lineage. In the Lineage event ‘Mirror War,’ AI enemies invade the server and control dungeons, which users must repel. The AI enemies with reinforcement learning models continuously learn through trial and error. They modify failed strategies and cooperate to attack users. The Mirror War event received positive feedback from users.

Ethical Use of Technology Considerations

NCSoft also deeply considers AI ethics. It continuously develops technologies to prevent AI from learning or generating unethical expressions such as profanity, prejudice, and disparagement. Technologies for personal data anonymization, prevention of derogatory expressions between countries, and recognition of unethical and hateful speech are applied in AI research and development.

For example, spam filtering technology is applied in chat systems of major games like Lineage 2M and Lineage W to identify and block various advertising patterns in 13 languages. Additionally, given the nature of games where users from multiple countries gather in one place, it prevents derogatory expressions between countries and controls the AI translation engine to identify and moderate expressions that could cause international disputes.

NCSoft is also striving to spread users’ awareness of AI ethics. A representative example is the dialogue series ‘AI Framework,’ where Chief Strategy Officer (CSO) Yoon Song-yi discussed AI ethics with world-renowned scholars. From April 2021 for one year, the AI dialogue content was published through official YouTube and blogs, featuring five global scholars from Harvard University, Stanford University, Massachusetts Institute of Technology (MIT), and others. The series was compiled into a book titled ‘The Most Human Future.’