King, the maker of Candy Crush, has long been leveraging AI and machine learning to enhance game development. This strategy predates their 2022 acquisition of the Swedish AI software company, Peltarion.
Under the leadership of Sahar Asadi, King’s AI Labs has developed AI tools that have significantly impacted games like Candy Crush. One notable tool is a playtesting bot that allows designers to understand player experiences before releasing levels, facilitating immediate refinements based on these insights.
The playtesting bot does more than just provide feedback; it can also automatically tweak levels to align with the designer’s intended player experience. Designers specify criteria, and the bot refines the levels, accordingly, presenting the best options for final approval.
This innovation reduces the time designers spend on mundane tasks, allowing them to focus more on the creative aspects of level design. Consequently, level tweaking has been sped up significantly due to AI intervention.
AI in gaming isn’t entirely new, as computers have played against human champions in games like chess for decades. However, King’s approach is distinct as their AI aims to emulate human players rather than outperform them.
The playtesting bot uses player data to mimic human-like behavior, predicting the actions a typical player might take at various stages of a game. This ensures that the game testing process is realistic and relevant to actual player experiences.
The bot’s ability to simulate human behavior is continually refined using the extensive data King collects from millions of player interactions. This allows the bot to predict the most human-like responses, even if they aren’t the optimal ones.
The bot’s performance is validated by comparing its level difficulty assessments with actual player experiences, ensuring it plays the game like a human. King is also working on incorporating player skills and preferences into the bot’s algorithms for even more accurate emulation.
The impact of this AI-driven approach has been profound. The playtesting bot has reduced manual tweaks to levels by 95% and sped up the overall tweaking process by 50%. Beyond speed, the bot also ensures the quality of released levels by assessing their playability, difficulty, and the balance of challenge they offer to players. These insights help designers make informed decisions about level releases.
Despite these advancements, there are concerns about AI potentially replacing human designers. However, Asadi emphasizes that designers are still crucial. The AI tools act as co-pilots, providing insights and assisting with mundane tasks, but the creative and fun aspects of game design remain in the hands of human designers.
The collaboration between designers and AI tools has been essential in developing these systems, and designers’ input is vital for creating enjoyable gameplay experiences.
The introduction of AI has changed the skill set required for game development roles, especially entry-level positions traditionally focused on manual tasks like testing. Asadi acknowledges this shift and highlights the evolving nature of job roles due to machine learning advancements.
While the foundational knowledge remains important, new technologies demand different skills and approaches, impacting how companies hire and train their employees.
Looking ahead, King’s AI Labs is exploring how AI can better understand player preferences and context to enhance gameplay experiences. By integrating foundational models and machine learning advancements, they aim to create more personalized and rewarding game interactions.
The rapid pace of AI development offers exciting opportunities for innovation, and King’s early investment in AI has positioned them to integrate these technologies swiftly into their games. For other companies, readiness to embrace AI’s potential will be crucial in leveraging these opportunities effectively.
