# World Model

#### **World Model for the Physical Reality**

The **World Model** is a digital representation for understanding the laws and phenomena of the physical world. It can be used to simulate and create digital worlds in a computational manner, while simultaneously enhancing the physical world itself. It represents an enhancement of how biological humans and bionic humans (robots), or AI-powered NPCs to interact with the world.

Human civilization was born in the physical world and has developed a profound capacity for understanding and creation. The anticipation of **physical** laws, phenomena, and development is a memory ingrained in human genes. We say something is **real** if that matches our natural experience as a biological human.

Therefore, even with the advancement of AI and computing, and even as **virtual** worlds and **digital** content become increasingly rich, greatly expanding human imagination, we still pursue a sense of "reality"—familiar laws, a tangible space we are part of, including visual, tactile, and spatial sensations that we hope are close to or consistent with that "**real**" world.

At the same time, the addition of the digital world has broadened the boundaries of the real physical world. This includes:

* **Augmenting Human Physical Boundaries** (augmented human)**:** For instance, seeing things invisible to the naked eye or touching non-existent objects in the world.
* **Embodied AI** (e.g., robots)**:** Reaching areas inaccessible to the human body.
* **Expanding the Boundaries of Human Imagination** (metaverse)**:** Creating virtual worlds that obey physical laws, generating digital content that doesn't exist in the physical world yet conforms to its rules.

All these are applications of different facets of the World Model, leading to varied emphases in understanding the world and implementing the digital model. However, these paths all adhere, at least superficially, to certain laws of the physical world—they are all "**realistic**."

Currently, there are three main application directions for so-called World Models:

1. **Realistic Video** (e.g., Sora)**:** Generates videos that follow physical world laws, looking "as real as the real thing," but which do not actually exist in the world. Such videos can be viewed directly, providing space for imagination, but they do not affect the physical world or how people perceive it. *(Video generation models like Sora, Veo et al.)*
2. **Realistic Digital Worlds** (e.g., Ready Player One)**:** Creates virtual digital worlds that obey physical laws, accessible via VR devices allowing interaction through controlled digital avatars—also "as real as the real thing." Such digital worlds can be used for games or creating better "realistic" videos, offering imaginative spaces and influencing human perception and interaction with the physical world, but without affecting the physical world itself. *(World generation models like those from Fei-Fei Li' World Labs et al.)*
3. **Realistic Physical AI:** Unlike the entirely digital products of the first two twin-world approaches, Physical AI involves a strong 1:1 binding between the physical world and the digital world that describes it. It emphasizes real-time, local, simultaneous, co-located fusion—highlighting immediacy, locality, immersion, and spatiotemporal consistency. It is also a twin of the world in real time, understood by machines (physical agents).  Virtual digital worlds are presented together within the same spatial coordinates and timeline of the physical real world, forming a hybrid model that merges one unique physical world with countless superimposed digital worlds. Each world exists in the same spacetime and adheres to the familiar physical laws—it's truly "**as real as the real thing"**, right here and right now. \
   \
   And when we say "**right here and right now**", we mean:\
       \-   Unlike a scanned digital twin 3D model from the real world, or even just a recorded video when you were there, you snapshotted the physical world and consume it sometime else and somewhere else (e.g. at your home's laptop).\
       \-   Unlike a AI-generated video or 3D virtual world space, it's in it's virtual timeline and parallel universe, and is a nonexistence in our physical world. Of cause you, as a human with physical spatiotemporal coordinates in our universe, do not share the same coordinates with those generated content, unless you also become a digital avatar (e.g. while playing VR games).

**Zeno's World Model** primarily serves this third path: **Realistic Physical AI**, and mutually embraces the generative world model paths of the first and second types, as you can put online content into a physical space. As mentioned, all these should accept humanity's genetic memory of physical laws. Furthermore, these paths may eventually merge into a comprehensive Grand World Model, encompassing capabilities for generation, creation, superimposing multiple realities, and controlling entities like robots in the real world.

Changes in the physical world directly affect the digital worlds, while tasks from digital worlds can also output back and control Physical AI agents in the physical world. For example, flipping a physical switch with an attached digital tag—it simultaneously impacts the real physical world, and vise versa.

#### Why Favor Physical Reality?

We build an enhanced model strongly bound to the real world, capable of understanding the spacetime coexisting with humans, augmenting human physical attributes with digital properties, and pulling people back to reality from consuming virtual immersion digital content endlessly, especially those AI generated. In short, we respecting **physical reality**, which will become a limited resource in the future technology. 

Even though the endless digital world and its content are fascinating and tempting, we, as human civilization, still reside in this vast, profound, and unique **physical universe**. Ultimately, our civilization will venture toward the distant stars and galaxies — those remote objects that our ancestors had been gazing since they were still Homo sapiens.