Economic Expectations and an AI Agent

The paper presents the cutting-edge approaches to economic expectations that originate in the field of computational intelligence. Firstly, the AI-based expectation concepts are situated within the long-standing debate about expectation formation in economics. They are discussed as a middle way between naïve and rational expectations and a subcategory of boundedly rational beliefs. Secondly, two approaches to formalize AI-based expectations (including their advantages and disadvantages) are presented in detail: One is based on Deep Reinforcement Learning and the other applies Recurrent Neural Networks. Finally, the theoretical implications of AI-based expectation concepts are discussed