Pixels and Profits: The future of AI-driven gaming markets

Abstract

Nvidia CEO Jensen Huang, in a recent Q&A session in March 2024, envisioned that video games with real-time content generated by AI are only 5-10 years away. This means that each of the 6-7 million polygons, a standard for modern AAA games, will be generated by deep neural networks 60 times a second on your screen, providing you with a novel experience each time you fire up your gaming device! The technology can prove to be disruptive to the 230-billion-dollar video game industry as generative game engines not only make each playthrough unique but also greatly reduce the game development workload, lowering cost and enabling game studios to focus on the narrative and creative design. While AI can render short videos, the computer resources and accuracy required to generate real-time game-level design and maintain it for the duration of an average gaming session without hallucinations seem unrealistic at best. However, technological innovations continue to surprise us. Days before writing this article, we find that the gaming community’s favorite ultra-violent video game - DOOM, rips and tears once again, with innovation as it becomes the first ‘first-person shooter’ trained on a generative diffusion model GamenGen (pronounced: game-engine). Developed by researchers at Google and Tel Aviv University, it is the first game engine powered entirely by a neural model that enables real-time interaction with a complex environment over long trajectories at high quality.