Video games will be the first to have minds controlled by generative AI.

• Inworld AI is a key example of Microsoft's big plans for generative AI in gaming.
• Non-playable characters, or NPCs, in video games are the focus of Inworld. These characters inhabit generated worlds but have been constrained by limited scripts.
• Inworld AI's chief product officer and co-founder, Kylan Gibbs, asserts that AI NPCs represent more than just a technological advancement; they symbolize a fundamental transformation in player engagement.

The rapid advance in generative artificial intelligence will not only transform human life but also revolutionize non-playable characters (NPCs) in video games. These characters, who have been limited in their abilities due to scripts, are now being tested as a core aspect of gaming where AI can enhance gameplay and immersiveness. A partnership between Xbox and Inworld AI is a prime example of this development.

Inworld AI's chief product officer and co-founder, Kylan Gibbs, stated that their technology enables NPCs to surpass predetermined roles, adjust to player actions, acquire from interactions, and contribute to a dynamic game environment. AI NPCs represent a revolutionary approach to player engagement, according to Gibbs.

Gaming companies and game developers have a significant chance to enhance immersion and boost replayability, retention, and revenue by transitioning from predetermined dialogue to player-driven narratives.

For years, the success story at has been driven by the interaction between powerful chips and gaming. However, there is now a growing sense in the gaming industry that AI is on the verge of taking off, despite some initial uncertainty.

John Spitzer, vice president of developer and performance technology at Nvidia, stated that all developers are interested in how artificial intelligence can affect the game development process, and he highlighted powering non-playable characters as a crucial test case.

The development of gaming worlds is limited by technological constraints, according to Gibbs. He believes that the advancements in AI NPC technology will pave the way for a new era of storytelling, creativity, and innovative gameplay. However, much of what is yet to come will be "games we have yet to imagine."

Bing Gordon, an Inworld advisor and former chief creative officer at Electronic Arts, stated that recent advancements in gaming have been driven by enhancements in visual fidelity and graphics. Gordon, who is now chief product officer at Kleiner Perkins and serves on the board of a gaming company, believes that AI will revolutionize the gaming industry for both players and game designers.

AI has the potential to create immersive worlds and complex narratives that center players in the fantasy, according to Gordon. Additionally, AI that impacts core game mechanics can boost engagement and draw players further into the game.

The potential for gen AI to revolutionize gaming production is the first significant opportunity, according to Anders Christofferson, a partner at Bain & Company's media & entertainment practice.

Gen AI is already doing days of work in minutes in professional tasks such as creating presentations and speeches. The initial storyboard design and NPC dialogue creation are handled by gen AI, allowing developers to focus on the more immersive and creative aspects of game making, according to Christofferson.

Creating unpredictable worlds

Bain's research indicates that AI is currently handling some tasks in game development, such as preproduction and planning. In the near future, AI will take on a larger role in developing characters, dialogue, and environments. Gaming executives expect AI to manage more than half of game development within the next five to ten years. While this may not result in lower production costs, blockbuster games can have total development costs of $1 billion. However, AI will enable games to be delivered more quickly and with improved quality.

The widespread use of gen AI will enable the average gamer to contribute to the creation of games through a "create mode," resulting in more user-generated content, as described by Christofferson as "player-driven narratives."

The shortage of human talent in the software engineering industry is a widespread issue that AI cannot solve quickly. However, it may provide developers with more time to focus on creative tasks and learn how to use new technology effectively as they experiment. A recent CNBC study found that 72% of workers who use AI reported increased productivity, which aligns with Microsoft's research on the impact of its Copilot AI in the workplace.

Christofferson stated that the gaming industry's use of GenAI is still in its early stages, and the landscape of players, services, etc. is constantly evolving. As with any new technology, there will be a lot of learning to take place in the coming years.

Julian Togelius, an associate professor of computer science and engineering at New York University, stated that the rapid changes in gaming may make it challenging to accurately predict the future of AI at this time. He characterized the current state of AI implementation as a "medium-size deal."

Togelius stated that generative AI is already being utilized by many individuals in the game development process. Programmers employ Copilot and ChatGPT to assist them in writing code, while concept artists utilize Stable Diffusion and Midjourney for experimentation. Additionally, there is a significant amount of interest in automated game testing and other forms of AI-enhanced QA.

The Microsoft and Inworld collaboration will examine the potential of AI in the video game industry, specifically in the areas of design-time and narrative generation. By utilizing AI to generate unique backstories for each of the thousands of NPCs in a game, developers can save significant time during the development process. Additionally, integrating generative AI into gameplay could further enhance the player experience.

"Getting this right is more challenging, as LLMs are known to cause hallucinations and games are not designed for this purpose," Togelius stated.

Togelius stated that video games are designed with predictable NPCs that adhere to a predetermined script and do not deviate from the game world by discussing non-existent topics.

The predictability of NPCs in video games is a core aspect of their design process, as it allows for a sense of control and reliability. However, open-ended games offer infinite possibility, which can be thrilling for players. Unpredictability in gaming is a new frontier, and could hinder the widespread use of AI. Finding the right balance between predictability and unpredictability will be crucial in moving forward with AI in gaming.

Togelius stated that modern AI will increasingly be incorporated into games and game development, necessitating new designs that take advantage of its strengths and limitations.