NCSOFT introduced results of its R&D on AI (Artificial Intelligence) technologies. It has secured highly applicable technologies such as ones that make development stages of games more effective and ones that intellectualize contents.
NCSOFT shared current status and activities of its R&D at AI Media Talk that was held in Pangyo on the 15th. It started relevant R&D on AI technologies since 2011 and it is currently carrying out its R&D with AI Center and NLP (Natural Language Processing) Center at the center.
NCSOFT secured AI-based technologies in five fields such as game AI, speech, vision (image and video), language AI, and intelligence AI and it is planning to focus its capabilities in innovating these five fields based on AI technologies. It already secured some technologies that can be commercialized, and it is operating a lab in each field.
◊NCSOFT Develops AI Technologies That Help with Game Production and Grow One’s Happiness
AI Center researches on solutions that help with game plays and development of games. “Our AI technologies are tools that solve problems.” said Director Lee Jae-joon of AI Center. “Our technologies will provide better solutions and new values and experiences to users.”
Game AI Lab develops development of games that are based on reinforcement learning and deep learning and service technologies. Based on reinforcement learning, deep learning, and simulation technologies, it researches on AI technologies such as AI technologies for game playing, AI technologies for game planning, and AI technologies for development of game art that are needed for development of games and services.
NCSOFT applied AI technologies to ‘Infinite Tower’, which is a content within its online game called ‘Blade & Soul’, and created an environment where users battle with AI. It is as if users battle with AlphaGo within a game.
Recently, it improved performance of its AI technologies through deep reinforcement learning that applies deep learning to current reinforcement learning. It is currently developing AI arena where AI is controlling battles just like how actual person is controlling battles by using battle logs of a user.
Speech Lab researches on technologies that recognize feelings and synthesizing voices. It researches on technologies that recognize languages, narrations, and feelings that are included in voice signals and technologies that that change texts into voices such as ones that are filled with dialogic styles and feelings. When these technologies are used, it is possible for gamers to share stories and enjoy game plays through voices that are synthesized according to characteristics of characters.
Vision TF (Task Force) researches on AI technologies related to image and video. Its AI technologies can recognize images or videos or create images by utilizing GAN (Generative Adversarial Network) technologies. They can either automatically provide tag information to graphic resources or color on their own and automatically create necessary images.

Photo Image
<‘NCSOFT AI Day 2018’ that was held at Pangyo R&D Center from the 22nd of February until the 23rd of February>

◊NCSOFT Plans to Expand AI Technologies to Other Fields
NLP Center focuses on creating an active ‘digital friend’, which can be applied to IT fields as well.
“One of our goals is to create an AI friend that provides information that may interest a user and answers to questions.” said Director Jang Jung-seon of NLP Center.
Besides natural language processing-based technologies, Language AI Lab researches on Q&A technologies, communication technologies, document summary technologies, and story creation technologies. These technologies allow AI to understand and summarize important information out of texts.
Intelligence AI Lab extracts and stores meaningful knowledge from various data such as logs or texts and it researches on technologies that deduce new knowledge or create and send new knowledge.
NLP Center introduced an intelligent baseball service called ‘NC PAIGE’, which took 3 years of research, on this day. It is a solution that creates story-based contents and communicates with users based on baseball data. It is sort of like a ‘baseball friend’ that cheers a same team as a user.
Staff Reporter Kim, Siso | siso@etnews.com