Featured Image for Article

AI-Generated Doom: A New Era in Game Development

AI-Generated Doom: A New Era in Game Development

Gamescom 2024 has brought to light some extraordinary advancements in video game technology. Among the highlights is a neural network capable of predicting gameplay at 20fps, a significant milestone in the realm of artificial intelligence.

Doom, the iconic first-person shooter, has been ported to numerous devices over the years. However, it has never been quite playable in the manner recently achieved by Google researchers. They have created an AI version of Doom entirely generated via a neural network. This innovative approach relies on video clips of gameplay to produce a version of the game that is nearly indistinguishable from the original.

An enhanced bundle of Doom 1 & 2 recently surfaced on Steam, featuring a new episode from MachineGames. This release coincides with the publication of a paper titled “Diffusion models are real-time game engines,” which documents the groundbreaking work of a small team from Google. The team has managed to interactively simulate a version of Doom using their game engine, GameNGen, powered entirely by a neural model.

GameNGen: The Future of Game Engines

GameNGen operates on the principle that AI can learn from and generate complex game states, much like how it generates static images from large datasets. By ingesting video clips from human gameplay, GameNGen produces sequential frames at 20fps with visual quality comparable to the original Doom.

The research paper poses an essential question: “Can a neural model running in real-time simulate a complex game at high quality?” The demonstrated answer is affirmative. The neural model can perform complex game state updates, such as tallying health and ammo, attacking enemies, damaging objects, and opening doors, maintaining game state over extended periods.

Implications and Future Prospects

GameNGen represents a significant step towards a new paradigm for game engines, one where games are automatically generated by neural models. This advancement raises critical questions about the training of these neural game engines and the role of human input in the creation process. Despite these challenges, the potential for AI-generated games is immense, and researchers are optimistic about the future.

Earlier this year, a Eurogamer investigation highlighted the transformative impact of AI on video game development. A report from Unity in March revealed that over 60% of game developers are incorporating AI at some stage in their development process, signaling a major shift in the industry.

Comments

No comments yet. Why don’t you start the discussion?

Leave a Reply

Your email address will not be published. Required fields are marked *