Palo Alto, California, October 16th, 2024, GamingWire
Tales—the popular AI gaming project spearheaded by a team of Stanford PhDs and Game developers—has today announced its innovative debut product: a Large World Model (LWM) that can generate entire digital worlds using simple text-to-game prompts.
Tales intends to completely transform the world of gaming by enabling users to create fully functional games without the need for any development or game design experience. The aim is to empower players to overthrow the gaming establishment.
Video: https://www.youtube.com/watch?v=_CUoVGW8VCw
Competing with the world’s leading studios, Tales allows gamers to ‘bring this vision to life’ by creating user-generated, interactive, and immersive experiences. By crowdsourcing data, resources, and infrastructure, Tales is developing Sia, perhaps the most radical AI model to-date.
Building games in seconds with a text prompt
Similar to how chatbots like ChatGPT and Claude are powered by Large Language Models (LLMs), which specialize in processing and generating text-based outputs, Tales is building its own Large World Model designed for video game creation. This LWM is capable of understanding and generating all components of a video game— from environments, physics, 3D models, and gameplay footage to NPC (non-player character) behavior—along with detailed metadata.
While an LLM learns language patterns by training on vast amounts of text, an LWM trains on gameplay data, video content, 3D assets, descriptive metadata, and feedback over time. This enables the LWM to understand how games are structured, refining its ability to create complex game elements and mechanics.
Tales can output fully functional games by leveraging 3D engines, spatial reasoning algorithms, and NPC behavior systems. With just a simple text prompt like, “make a first person shooter in space” the LWM can generate the requested game, which is customizable and immediately playable. The model essentially serves as a tool for procedurally generating games in real-time and marks a substantial evolution from existing AI offerings.
Tales’ outputs extend beyond games, with possibilities for virtual reality worlds, interactive experiences, and immersive educational tools, making this a diversified value proposition for potential future investors.
How does Tales learn?
In one of the most ambitious crowdsourced projects in history, Tales utilizes data, storage, and computational power to train its Large World Model. To ensure the community are engaged and incentivized to build this together, Tales is launching a rewards-based incentive system.
Users can submit anything from gameplay footage, in-game assets, and extensive environment descriptions to form a fully transparent dataset, which is tracked and documented to ensure provenance and preserve privacy.
Tales will be ready for early access in November. For more information, please visit https://tales.world/.
About Tales
Tales is an ambitious next-generation gaming platform, powered by AI. Tales is pioneering crowdsourced generative AI within the gaming industry with its large world model. By crowdsourcing vast amounts of data and computational power, the team including Stanford PhDs and Game Developers based out of Palo Alto, are building one of the world’s most powerful AI models ever, democratizing game creation and empowering creators to bring their vision to life with zero friction.
To follow on social media: X, Telegram, Discord, TikTok, Twitch, Medium, YouTube
Contact
Head of Community
Viktor Uzunov
Tales
[email protected]
Disclaimer: Press release sponsored by our commercial partners.
Daily Debrief Newsletter
Start every day with the top news stories right now, plus original features, a podcast, videos and more.
Source: https://decrypt.co/286607/tales-reveals-gaming-ai-enables-users-to-build-play-any-game