Mixed-Initiative Level Design with RL Brush
This addresses the challenge of enhancing creativity and efficiency in game level design for designers and developers, though it is incremental as it builds on existing mixed-initiative and RL approaches.
The paper tackles the problem of level design for tile-based games by introducing RL Brush, a mixed-initiative tool that uses reinforcement learning to provide AI-generated suggestions, resulting in users creating more playable and complex levels in Sokoban, with tests showing increased engagement and improved level quality in 39 sessions.
This paper introduces RL Brush, a level-editing tool for tile-based games designed for mixed-initiative co-creation. The tool uses reinforcement-learning-based models to augment manual human level-design through the addition of AI-generated suggestions. Here, we apply RL Brush to designing levels for the classic puzzle game Sokoban. We put the tool online and tested it in 39 different sessions. The results show that users using the AI suggestions stay around longer and their created levels on average are more playable and more complex than without.