Summary of Understanding goals in complex systems

00:00:00 - 00:10:00

This video discusses the goals of complex systems, and how to achieve them. One experimental idea is to induce local goals by fitting a neural network to a task and its goals. The project is to figure out how goals are encoded in the information that flows between goals and perceptions.

• 00:00:00 The author introduces the idea of goals in complex systems, explaining that these goals can be embodied in a neural network as weights and that they can be used to ask questions about computational systems. He goes on to introduce the concept of the locus, which is a box that holds all the possible configurations of parameters that can be used to induce a meaningful behavior. He argues that the power of the locus can be used to ask questions about deep learning and the interface of a computer program like Loki.
• 00:05:00 The video discusses the goal of a complex system, which is to achieve a desired outcome. A modulator is a component of a Locus that helps the system achieve this goal. The modulator can be thought of as a concrete stack of assignments that help the system achieve its desired outcome.
• 00:10:00 This video discusses the goals of complex systems, and how to achieve them. One experimental idea is to induce local goals by fitting a neural network to a task and its goals. The project is to figure out how goals are encoded in the information that flows between goals and perceptions.