Closing the Designer / Player Gap
There is a lot of noise about what Generative AI will mean for video games. The predominant conversations have been around faster pipelines and production, with the most ghoulish of said discussions focused on efforts to replace human creatives (a topic Yichao also reflected on late last year).
I'd like to highlight one innovation that I haven't yet seen discussed broadly: how AI enables us to close the feedback loop between player and designer, unlocking brand new kinds of games.
Games have long chased closing the gap between players and adaptive and additional content from designers. Earlier in the industry, when games were delivered via CD’s and cartridges, there was a complete separation between designer input on content (locked on the shipping of the game) and when players experienced it. As games evolved to digital platforms, expansions and downloadable content (DLC) enabled designers to provide updates and responses to players—to improve the game by listening to the community of players. Nowadays, in live service games, patch and update cadences give designers unprecedented ability to react to players, listen to their community, and update content in response to player actions and needs.
Another aspect of closing this loop in games is crafting an experience that feels like the game is adapting and responding to the players’ actions. An aspirational example for many digital games is the tabletop RPG Dungeon Master. In tabletop, the DM acts as the designer of a fantastical setting that plays with the players, tuning and adjusting the story and experience in real time during the play sessions.
Earlier classical AI Directors pioneered by games like Left 4 Dead were explorations in this vein—automated, adaptive algorithms that take designer intent (of an experience filled with action, tension, and danger) and adjusts the difficulty, tempo, and cadence of the designed experience based on the choices and actions the players make. But these classical AI systems could only ever be impartial umpires of game rules, setting the stage and executing the logic without actively particpating or adjusting the narrative.
Generative AI, when embedded at the heart of a game engine and play loop, changes all of that.
Rather than designer intent only getting injected into game experiences on a bi-weekly (or longer) cadence via updates, GenAI game systems can be crafted by devs to adapt the designer-created content to player actions in real time. And like a live DM, games powered by GenAI can respond and tune difficulty and the story on the fly, reacting dynamically to player creativity and choices.
The potential here is not the replacement of a designer, but rather the extension of their creativity and intentions into the game system itself. By leveraging machine learning, a game can now take designed content and adapt it to player actions with a level of dynamism previously unattainable. This approach allows for the system to obtain a nuanced understanding of player behavior and player intention, leading to a play experience that is both personalized and deeply immersive.
By placing an AI proxy of the designer within the core game loop, we are witnessing the dawn of a new era in gaming—an era where the boundaries between creation and interaction are seamlessly intertwined, offering unparalleled gaming experiences that are as dynamic as they are engaging.
In the coming weeks, we’re excited to share peeks at how we’re doing this with Project Emily—our flagship game. Keep an eye on our socials across TikTok, Instagram, LinkedIn, and more for insights and behind the scenes looks. We’ll also be sharing a technical demo at GDC with a select group of devs and interested orgs. If you’d like to be considered for this early look, drop us a line at press@jamandtea.studio.
