Support Autonomy: Exploring Player Perspectives on AI-Supported Onboarding in Video Games

Research Background and Issues

• Identified Problems and Challenges:

  • During the game onboarding process, it is essential to ensure that players can learn game mechanics in an engaging and participatory manner. However, traditional onboarding methods (such as tutorials and text instructions) are often perceived as lengthy, unappealing, and difficult to personalize. Players also have diverse learning needs.
  • Excessive information may lead to cognitive overload, causing players to feel frustrated in the early stages and abandon the game.
  • There is currently limited research on how artificial intelligence (AI) can optimize this onboarding process, particularly in terms of qualitative studies on player experience.

• Significance:

  • The onboarding process is critical to players' first impressions and long-term retention, especially in free-to-play service-based games and multiplayer games that require an active player base.
  • A well-designed onboarding experience not only helps players quickly get started but also effectively extends their game lifecycle.

• Research Motivation and Related Work:

  • Many games use AI to simulate non-player characters (NPCs), generate content, or manage player data, demonstrating AI's potential in enhancing engagement and personalization.
  • AI can dynamically adapt to player behavior, providing personalized suggestions or adjusting opponent behavior, thereby creating a richer and more efficient learning experience during onboarding.

Proposed Solution

• Proposed Solution:

  • The study designed an AI-based recommendation system for game onboarding, focusing on the impact of providing real-time personalized guidance on player experience during the initial stages.
  • By developing an original mobile game, Joker, the study compared two onboarding conditions: player self-guided (without AI involvement) and AI-assisted (AI providing optional suggestions each round).

• Innovations:

  • This is the first systematic study from the player's perspective on the impact of AI-guided onboarding on individualized experiences and learning processes, rather than solely focusing on technical implementation.
  • A teaching-assistive AI system was designed and implemented, which analyzes players' actions in real-time and provides optional suggestions, making the onboarding process more dynamic and responsive while maintaining players' control over the learning process.

• Implementation Steps:

  1. Design an original turn-based strategy game, Joker, combining familiar elements (e.g., game boards and playing cards) with complex new rules.
  2. In the game, the AI algorithm uses a decision tree method to provide real-time suggestions based on players' cards and game situations, such as optimal card combinations or best layouts.
  3. Collect data through two rounds of gameplay with 20 participants under self-guided and AI-assisted conditions, followed by qualitative thematic analysis.

Research Findings

• Key Findings:

  • Player Learning Preferences: Players preferred to learn through actual gameplay (e.g., trial and error or repeated practice) rather than relying solely on text-based tutorials.
  • Player Expectations of AI:
    1. Personalization: Players expect AI to dynamically adapt to their individual learning progress and needs, addressing the shortcomings of "one-size-fits-all" onboarding.
    2. Transparency: Players want AI not only to provide suggestions but also to explain the logic behind them to enhance trust.
    3. Autonomy: AI should be an optional feature rather than a mandatory component, as players desire full control over the onboarding process.
  • AI's Actual Impact on Learning: While AI can reduce cognitive load and alleviate early-stage difficulties, mechanically following AI suggestions may weaken players' active learning and long-term mastery.

• Advantages Over Existing Solutions:

  • Proposed the "information transparency + free choice" AI design principle, balancing technological assistance with player autonomy.
  • Explored how AI-guided onboarding can dynamically intervene in players' learning behaviors rather than relying on static tutorials.

• Experimental or Evaluation Results:

  • Players generally had a positive impression of the AI, particularly its ability to reduce frustration caused by aimless exploration. Among the 20 participants, 12 preferred the AI-assisted condition.
  • However, heavy reliance on AI suggestions led to players losing in some scenarios, as those who strictly followed AI recommendations struggled to independently develop strategies or learn the rules.
  • Data also showed that when AI assistance was used in the first round, players had a lower win rate in the second round under non-AI conditions, potentially reflecting limitations in learning outcomes.

• Limitations and Future Research Directions:

  • Limitations: Most participants were highly educated players, which may not represent a broader player demographic. The onboarding method in Joker primarily explored real-time suggestion strategies, excluding other AI-guided forms (e.g., adaptive tutorials).
  • Future Directions:
    1. Develop more advanced machine learning algorithms to achieve more personalized suggestions.
    2. Explore AI applications in other types of onboarding, such as dynamic opponent matching or game environment generation.
    3. Conduct longitudinal studies to observe the impact of AI-guided onboarding on players' long-term skill acquisition and engagement.

In summary, this study highlights the potential of AI-guided onboarding in games while emphasizing the importance of maintaining player agency. These findings provide valuable insights for designing more player-centric AI game systems in the future.