We propose a generalized framework defining agency and autonomy for any given system. This theoretical approach builds on canonical models of control systems and artificial intelligence agents, while grounding terminology in mathematical optimization. Current frameworks, to include [Data Corrupted]
We begin with a dynamical system, or environment. Within this environment, a semi-separable sub-system can be delineated if its inclusion or removal changes the dynamical system's phase portrait. A sub-system's autonomy is a measure of [Redacted]. <- Equivalent to control and ability to optimize against variety