COMP_SCI 496: AI Perspectives: Symbolic Reasoning to Deep Learning



See prerequisites below.


Prerequisites: This class is by application only, as instructor permission is required---possibly based on a brief meeting---in addition to (CS 349 AND (CS 348 or CS 371)). Equivalent courses elsewhere may be considered if course descriptions are provided and approved, though students who have taken the prerequisites at Northwestern will be given priority.

This is an interactive course designed to engage students in incisive and energetic discussions on different perspectives in Artificial Intelligence (AI), from classical AI approaches to recent advances in machine learning. The course will feature guest lectures by Northwestern professors from various disciplines and external guest speakers, both offering their takes on AI: comparing knowledge representation and symbolic reasoning with the state of the art in deep learning, exploring cognition through psychology, neuroscience, and linguistics, looking at consciousness through the lens of philosophy, unpacking the ethics of AI, and more.

Note that this course does not teach the how-to’s of either symbolic reasoning or deep learning, i.e., this course does not teach algorithms and methods in AI; rather, it discusses various aspects of symbolic reasoning, deep learning, and everything in between: how we do things, why we do them, the meaning of it all, the benefits and dangers of what we do, and where may go from here.

Coursework will center on class participation: students will read research papers, view debates and debate views, and share their evolving perspectives by writing weekly thought pieces and a final paper, which may turn into a final presentation instead. However, the course components and expectations may evolve and significantly change with the progression of class conversations.

Grading in this course expects heavy class participation involving a lot of talking and a significant amount of reading and writing. There is no coding, nor exams, but there may be pop quizzes. In-person class attendance is mandatory, and exceptions will only be made at the discretion of the instructor for urgent situations such as health issues or other emergencies that are truly outside of the student’s control.

REQUIRED TEXTBOOK: This course doesn’t use any textbooks, but topics and aspects from the following books have come up in discussion in the past and are recommended as pre-reading prior to the start of the course:
1. The Wisdom of Crowds by James Surowiecki
2. Superintelligence by Nick Bostrom
3. Ant Encounters: Interaction Networks and Colony Behavior by Deborah M. Gordon
4. Deep Thinking by Grandmaster Garry Kasparov



COURSE GOALS: The goal of the course is to give students a holistic view of the field of AI and to help them to ask and answer important questions in their own academic or industry pursuits in AI: Will intelligence be an emergent property of neural networks? Should cognitive architectures utilize knowledge representations instead? What would a hybrid system look like that combines the two approaches? Is there a novel approach that departs altogether from the familiar? Are current large language models the final answer to our quest for AI? The course expects to invigorate students with an enthusiasm about AI that can enable them to move the field forward.