Speakers

Yeonu Jeong is a game data analyst at Nexon Korea, where he leads analysis of how players perceive in-game AI features. He defined an eight-dimension taxonomy of how players perceive in-game AI, applied it to player reviews across 172 Steam titles, and used logistic regression to estimate how each dimension affects whether players recommend a game. He developed the results into a pre-launch diagnostic tool that produces an acceptance score and dimension-level risk factors for planned AI features, and validated it on shipped titles. He previously handled LTV model operations and root cause analysis of metric movements for live games. During his MS in Statistics at Korea University, he developed a penalized double-layer neural network regression model that detects underlying structure in data and presented it at the Korean Statistical Society conference; that statistical modeling background informed the regression analysis behind his in-game AI research. In this session he examines which in-game AI players welcome and which they reject, and how to check for that before launch.

Presenting:

• Game AI Through Players' Eyes: What Works, What Backfires