Closed SeverTopan closed 5 years ago
A System would then be categorized as an Actor, or an Agent based on the (Kolmogorov?) complexity of the function correlating its Decision (#43).
There would then be a gradient between Actor and Agent, with the delineation drawn arbitrarily.
Maybe an Agent is an Actor that uses a World Model?
I would think Actors are Systems rather than the other way around but I'm willing to be convinced...
I agree that Actors are Systems, I think my comment was a little ambiguous. What I was trying to say is that the assessment of whether something is an Agent or an Actor relies on the Agent or Actor's Decision complexity.
Re: https://github.com/Vulcan-Academy/Vulcan-Academy.github.io/issues/40#issuecomment-373399012, I'd argue the presence of a World Model factors into Decision complexity.
I agree the World Model factors into Decision complexity, but I would say that distinguishes simple agents from complex agents rather than actors from agents. To test this design how would you classify gravity?
What if i have a very simple world model? Conversely, we can imagine a very complex Decision that does not utilize a World Model at all.
My interpretation is that a World Model is just a strategy with which we can increase the complexity of a Decision, but its not strictly needed for a complex agent. The World Model is in a way anthropocentric.
Gravity could be defined as an Action of the "Physics" Actor. There are multiple ways to represent it but I think it sits at the low Decision Complexity end of the spectrum.
An Actor is the base class of an Agent. It can perform actions, but does not exhibit philosophical agency. E.g. A thermostat can be defined as an Actor. This should resolve the agency ambiguity in #28.