We demonstrate how users can create and manipulate causal virtual worlds using large‑language‑model (LLM) and large‑video‑model agents. The approach relies on feedback fuzzy cognitive maps (FCMs) to model the granular causal structure of the virtual world and to steer its causal evolution. Local causal rules are partial or fuzzy, while the feedback structure of the FCM yields global equilibria that define causal scenarios.
A sequence of dynamical meta‑rules of the form "If $\mathcal{A}$ then $\mathcal{B}$" defines the video’s causal scenes. The if‑part $\mathcal{A}$ perturbs the FCM at the discretion of the user or agent; the transient feedback dynamics of the FCM determine the implication arrow of the meta‑rule; the then‑part $\mathcal{B}$ manifests as an equilibrium attractor such as an FCM limit cycle or fixed point.
Our algorithm extracts these meta‑rules from the FCM and guides the LLM agent to write a script based on the meta‑rule sequence. The large video generator then converts the script into a video scene consistent with the dynamics flow.
In a case study we applied the technique to a simple FCM describing an undersea world of dolphins and sharks. Google Gemini 3.1 generated the script, and Google Veo 3.1 produced the dolphin‑shark video. The approach is general and can scale by combining larger FCMs and AI agents to create more immersive virtual worlds.
Review