Related Experiment Video
Updated: Aug 5, 2026
03:14
Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
Published on: December 6, 2024
Adaptive level modification via player skill classification and large language models
Ahmed A Elshamy1, Hazem N Aliedin1, Shehab T Shaban1
1Department of Computer Science and Information Technology, Egypt-Japan University of Science and Technology (E-JUST), New Borg El Arab City, Alexandria, Egypt.
Scientific Reports
|July 28, 2026
Summary
This study introduces an adaptive framework that modifies video game levels in real-time to match player skill. This personalized approach enhances player engagement by adjusting game difficulty dynamically.
Area of Science:
- Artificial Intelligence
- Game Design
- Human-Computer Interaction
Background:
- Player engagement in video games hinges on balancing challenge and competence.
- Current dynamic difficulty adjustment systems lack structural level modification capabilities.
- Static difficulty settings do not accommodate individual player skill variations.
Purpose of the Study:
- To present an adaptive level modification framework for personalized video game experiences.
- To infer player skill in real-time and structurally modify game levels accordingly.
- To improve player engagement through tailored gameplay challenges.
Main Methods:
- Constructed a hybrid dataset using Proximal Policy Optimization (PPO) agent trajectories and human gameplay data.
- Developed a classifier achieving 97.82% accuracy to categorize players into expert, normal, and beginner skill levels.
- Utilized a two-stage large language model (LLM) pipeline for real-time, skill-conditioned level modifications, verified for traversability.
Main Results:
- The framework demonstrated high accuracy in classifying player skill levels.
- Modified Super Mario Bros. levels achieved a 74.1% playability rate at full-level granularity and 83.5% at chunk granularity.
- The system successfully adapted level structures to maintain appropriate challenge.
Conclusions:
- The adaptive level modification framework offers a novel approach to dynamic difficulty adjustment.
- Real-time structural modifications enhance personalized gameplay and player engagement.
- The developed methods show promise for creating more adaptive and engaging video game experiences.
