Emergence of Implicit World Models from Mortal Agents
This is a theoretical discussion with no empirical results, so it is incremental and speculative, addressing foundational questions in AI and autonomous systems.
The paper explores how world models and active exploration might emerge implicitly in autonomous agents through open-ended behavior optimization, focusing on homeostasis as a key objective and proposing a hypothetical architecture combining meta-reinforcement learning for robust adaptation.
We discuss the possibility of world models and active exploration as emergent properties of open-ended behavior optimization in autonomous agents. In discussing the source of the open-endedness of living things, we start from the perspective of biological systems as understood by the mechanistic approach of theoretical biology and artificial life. From this perspective, we discuss the potential of homeostasis in particular as an open-ended objective for autonomous agents and as a general, integrative extrinsic motivation. We then discuss the possibility of implicitly acquiring a world model and active exploration through the internal dynamics of a network, and a hypothetical architecture for this, by combining meta-reinforcement learning, which assumes domain adaptation as a system that achieves robust homeostasis.