Me & You vs. AI – Enhancing Cooperative Gameplay and Conversation through Generative AI
Hessische Abschlussförderung - Master Thesis by Mayuri Sajnani
Ein Beitrag von Mayuri Sajnani
Dienstag, 9. April 2024
Mediencampus der Hochschule Darmstadt
The Master’s thesis project „Me and You versus AI“ is an experimental approach to game design, casting generative AI and NLP in the role of a mythological being — an „AI god“— guiding, mediating, and conversing with two players while controlling the game world around them. Inspired by the myths and traditions of Hinduism, such as the custom of scattering the ashes of the deceased in the River Ganges, this project brings a new twist to how we interact with AI. Not alone, but through a cooperative journey testing our human relationships.
It features a melancholic tale of two estranged siblings on a quest to release their parents‘ ashes at a river temple, only to be challenged by an embodiment of the river itself, a charismatic river deity driven by AI, with a vengeful agenda of her own. She has submerged the city and the temple under water, presenting the siblings with trials. Under the pretense of aiding them, she tests their resolve to reach the temple’s heart.
In our modern world, where faith and technology often clash, AI emerges as both an object of fascination and a reflection of our society’s complexities. This research re-imagines AI as a companion on the edge of deism and humanity. „Me and You versus AI“ invites us to interact with AI while scrutinizing its motives.
This project introduces a novel cooperative gaming experience powered by advanced natural language processing (NLP), enabling real-time dialogue and feedback for players, where AI is not just for chatting but integrated as a fundamental part of the game.
The game design is tested through a prototype developed using a complex game architecture featuring speech recognition, text and image generation and speech synthesis, all working seamlessly in real time. The gameplay focuses on puzzles themed around relationship-building techniques, and through the integration of AI, it adapts the puzzles to match the players‘ interests and intentions, making every playthrough unique and personalized.
Through this project, we also highlight the indispensable role of human developers in crafting interactive spaces with AI, utilizing the judgment and analytical capabilities of LLMs along with human production and creativity, to innovate human-AI interactions and even gamify them.