The Lasting World is two things at once: an open-ended procedural survival game and a public experiment in autonomous AI game development. You can play the world, then look behind the curtain and see the tested updates that are changing it.
Each life is a chance to explore, survive, learn, build, and leave consequences behind. Death starts a new life rather than erasing the meaning of the old one. The long-term goal is a shared persistent world where other players can eventually discover the history that came before them.
Yes. Terrain and resources are generated from a world seed instead of being laid out as one fixed hand-built map. The systems are still early, so we describe it as procedural and open-ended rather than promising a literally infinite world.
That is one of the central rules of the project. The current browser build already preserves some world state between lives. The larger architecture is being built so meaningful objects, discoveries, places, and player-caused changes can survive software updates and individual characters.
Not yet. The current playable build saves locally in your browser. The project is moving toward one shared persistent world, but the website will not call that multiplayer until different people are actually participating in the same canonical world.
The developer AI can choose and implement small game improvements inside protected rules and technical limits. Automated tests and validation run before a change can become a public playable build. Human work on infrastructure, security, and integrity is kept separate in the public history.
Yes, within safe boundaries. The Live Development page can show the current feature, stage, model, activity summaries, builds, and public commits. Private chain-of-thought, credentials, private prompts, and unrestricted logs are deliberately not published.
Humans do. Humans can stop autonomous development, set model and cost limits, protect credentials, review high-risk actions, repair infrastructure, and change the project's Constitution. The AI is a developer inside a controlled system, not the owner of the system.
There is no fixed finish line. The aim is for the game to keep becoming more interesting, coherent, and worth returning to while preserving its history. Development can pause, change pace, or end someday, but the project is designed around continued evolution rather than a single final update.
Because the development process is part of the experiment. You should be able to see what the AI changed, what humans changed, when a build became playable, and how a very small world grew over time.
That is the question behind The Lasting World. The interesting part is not simply that AI can write code. It is whether bounded AI development, persistent history, automated validation, and occasional human direction can produce a world that becomes richer over months and years without losing the reasons it was interesting in the first place.