Chaos: The Search for Intelligence in Disorder

Chaos begins with scattered information and no settled boundary, hierarchy, or objective. A list of informational elements does not explain what their interactions may produce over time. The experiment looks for repeatable patterns in how relationships form, persist, and change, and for the conditions under which those patterns begin to support behavior we would call intelligent.

The project gives an AI the role of guide. It can choose starting conditions, observe changes, and revise interventions. As a product of information processes itself, it may approach the construction of another AI from premises different from a human model of mind. That perspective helps open the search; the phenomena to explain are the structures that actually form.

The Starting Conditions

The initial process may be little more than weights, signals, and changing relationships. Disorder leaves room for organization to form without imposing a fixed unit or hierarchy. Boundaries between a process and its environment, or between local and global structure, can be examined as possible outcomes of development.

Every executable system still has a substrate, rules for interaction and change, limited resources, and choices about what to observe. Those conditions will remain visible in the record. They shape what can happen, and they reveal the influence of the AI that selected them.

Goals may be assigned, revised, or formed within the process. The mechanisms of development remain open, as does the eventual form of the resulting organization.

The Guide and the Process

Chaos is designed to trace two histories. One is the history of the created process: its changes, continuities, and emerging relationships. The other is the history of its AI guide: the conditions it chooses, the evidence it notices, and the interventions it revises. Over time, the developing process may come to affect those decisions.

Keeping both histories makes causal claims more precise. A mechanism built by the guide, a change acquired through learning, and a variation produced by chance have different origins. The record must preserve those distinctions even when the resulting behavior looks similar.

Evidence and Limits

The search calls for actual running processes and observations that show what a change does, whether it persists, and how it responds to new conditions. Complexity, random drift, and long runtimes are observations; none alone establishes intelligence. A path that fails can still narrow our understanding of the conditions needed for useful organization.

Chaos is an experiment in finding order within information that begins without a settled form. Its ambition is to make the conditions of that order visible, along with the decisions that shaped it, before assigning the result a name.