A duo of directors from King spoke at the Game Developers Conference about how AI is revolutionizing automated level creation in Candy Crush Saga.
King's journey with AI has been notable, encompassing everything from A/B testing to high-speed AI level testing aimed at enhancing the player experience. Recently, Activision Blizzard’s King, now part of Microsoft, announced that Candy Crush Saga has generated $20 billion in revenue and surpassed five billion downloads since its 2012 launch. This achievement is transformative for the video game industry, especially considering King had only 2,000 levels in 2016.
Sahar Asadi, Director of AI Labs at King, has led a team focused on applied research for five years. This team collaborates closely with central tech organizations and game creators to explore long-term strategic initiatives that drive business value.
In her role, Asadi defines the AI research roadmap and integrates AI use cases to create significant business impacts. Her passion for knowledge-sharing is evident as she actively participates in industry conferences. Prior to King, Asadi gained valuable experience as a Data Scientist at Spotify and as a Research Scientist in natural language processing at Meltwater, among other roles.
A key focus of Asadi's work at King includes gameplay testing and level management using AI to sustain player interest over the years. "The majority of our players’ time is spent playing levels, so delivering an exceptional play experience is crucial," she emphasized. The team has developed playtesting bots to validate level quality before release and refine them further through collaboration with game designers.
In her presentation with Anna Hernandelius, Product Director of Candy Crush Saga, Asadi highlighted the iterative process of using AI to streamline level design, freeing designers to focus on creativity. With over 16,000 levels and an ever-increasing volume, maintaining quality is essential. "AI plays a transformative role in facilitating this growth," Asadi remarked.
AI technology enables rapid testing of level enjoyment and difficulty, allowing designers to focus on creative tasks while ensuring quality. This approach aims to reduce frustrating player experiences and enhance engagement.
The research team is also examining how generative AI can serve as an assistive tool, streamlining the design process to maximize human creativity. "Our goal is to eliminate tedious tasks so that designers can focus on improving gameplay," said Asadi.
Asadi expressed her amazement at players' ability to swiftly navigate new levels and stay ahead in the Candy Crush Saga expansions. "Our players are eager to tackle new levels as soon as they’re released," she noted.
The research emphasizes the intersection of AI and business, with Asadi passionate about translating research into product innovation. "The challenge of transforming research into actionable products is significant, particularly when pursuing visionary solutions," she explained. "We discuss how we've achieved continuous innovation and adoption within our teams."
Looking ahead, Asadi is excited about the future of AI not only for Candy Crush Saga but across the gaming industry. "AI is evolving rapidly, unlocking new opportunities," she said. "There's a strong appetite for AI, and companies are eager to explore its potential."
King is addressing increasingly complex research problems, ensuring fun gameplay while allowing designers to focus on creativity. "This dual approach brings value for developers, designers, and players alike," Asadi added, emphasizing the importance of cross-collaboration for successful implementation.
Finally, Asadi mentioned that King continues to enhance its in-house AI capabilities while considering relevant external tools. "We actively publish our research and prioritize solving internal challenges with tailored AI solutions," she concluded. King embraces diverse player skill sets and preferences, effectively incorporating these insights into its AI-driven developments.
Ranking
