Fall 2024 Course Descriptions
CS 150-02 Cognitive Architectures Age of Foundation Models
This course will delve into the changes to cognitive architectures prompted by the rise and success of foundation models, in particular, large language models. Part of the goal of cognitive architecture was to capture human cognition and intelligence. Arguably, foundation models have made more progress on that end, at least on the intelligence part, than any of the past cognitive architecture were able to achieve. Yet, foundation models are not cognitively or neurally plausible, and they have a host of problems of their own. The course will review classical cognitive architecture and discuss ways for utilizing foundation models to improve them, e.g., by replace components of classical cognitive architecture with foundation models. The course will provide opportunities for students to computationally investigate different integration methods and their design tradeoffs in embodied and disembodied computational models.
Prerequisite: Solid programming knowledge in Java and Python and willingness to learn new representations and programming paradigms (e.g., production systems in classical cognitive architectures), knowledge in AI, robotics, algorithms, data structures, statistical and experimental methods.
Open to CS grad students and advanced CS UGs and CBS students with excellent programming skills and the ability to work independently on programming assignments