# FAQ

## Is [Brownyx](https://brownyx.com) CEL a new AI model?

No. CEL is an architecture layer, not a model.

It can be implemented using LLM calls, deterministic checks, retrieval, agenda systems, and validation logic.

## Is CEL the same as bisociation?

No. Bisociation is only one late-stage strategy inside CEL.

CEL also includes context sufficiency, intent clarification, frame switching, contradiction detection, hypothesis generation, self-model tension detection, question generation, analogy, and synthesis.

## Does CEL prove machine consciousness?

No.

CEL is an engineering pattern for bounded cognitive expansion. It does not prove consciousness.

## Why not run CEL on every message?

Because that would make the system expensive, noisy, and potentially obscure.

CEL should run only when direct reasoning is insufficient.

## Why is bisociation internal by default?

Because distant associations can easily become decorative or misleading.

A bisociation should influence user-facing output only when it produces a clear useful conclusion.

## Why avoid new memory types?

Because memory pollution is dangerous for persistent systems.

[Brownyx](https://brownyx.com) should store different cognitive products in the correct state layers: memory, questions, hypotheses, contradictions, MindCells, artifacts, and traces.

## Can other people implement CEL?

This repository documents the concept publicly. See LICENSE.md for documentation rights and NOTICE.md for reserved rights.

The production [Brownyx](https://brownyx.com) implementation is not included.

## Can CEL be used outside [Brownyx](https://brownyx.com)?

Conceptually, yes. CEL is described as a general architecture pattern, but the exact implementation here is [Brownyx](https://brownyx.com)-oriented.