AI Teammates in Overwatch 2: Investigating Player Experience and Game Dynamics through Self-Determination Theory

Sammanfattning

Artificial Intelligence (AI) has become an essential aid in making videogames. AI characters could have an impact on shaping players' experience, particularly in team-based multiplayer environments. In competitive environments, when players leave mid-match, it puts their team at a significant disadvantage, possibly disrupting balance and negatively impacting the overall gameplay experience. Valve’s earlier solution in Counter-Strike: Global Offensive (CS: GO), which employed semi-responsive AI to replace disconnected players, was discontinued in 2021. Since then, no viable alternatives have been introduced, leaving a critical gap in how games address team imbalance and player experience during incomplete matches. This study investigates how the integration of AI can be optimised to preserve or enhance the gaming experience for players, using Self-Determination Theory as a lens and Overwatch 2 as a case study. A mixed-methods user study was conducted using questionnaires, gameplay observation, and semi-structured focus groups. Findings revealed that AI teammates often undermine autonomy and competence due to rigid, context-unaware behaviours. Players reported communication breakdowns and a lack of strategic coordination, reducing their sense of agency and effectiveness. Trust, as a facet of relatedness, varied by context; players expressed more confidence in AI on attack than on defence. Players also reported diminished immersion, a key dimension of the lived experience explored through phenomenology. The findings reveal psychological and social tensions when AI replaces human players and suggest design implications for building more responsive, human-like AI teammates.