Are LLMs the path to AGI? Is superintelligence inevitable? How do we address the risks of AI? (including existential risk, deepfakes, impact on jobs, privacy, democracy) While neuroscience, philosophy, and psychology all provide insights into these questions, this course will focus on the Big Ideas drawn from the last 60+ years of AI research. We will seek to understand the foundations of machine learning (supervised, unsupervised, and self-supervised), the power of scale up, and other Big Ideas that will shape the field in the future.

Disclaimer: this course is being offered for the third time. It’s very different from previous offerings by a different instructor.

General information

• Instructor: Oren Etzioni, Email: etzioni@cs.washington.edu
• TA: Alexander Chin, Email: alexchin@cs.washington.edu
• Time: Tuesdays, 3:30 PM - 4:50 PM
• Location: CSE2 Room G04
• Ed board: here.

Expectations

1. Complete the readings before class.
2. Post writing assignments on the discussion board (use of a “GPT” is completely ok).
3. Be physically and mentally present in all class meetings (can miss at most 1).
4. Contribute to class discussions (including follow up on the discussion board).
5. No screens (phone or laptop) in class.

Topics/questions (evolving)

1. AI, The Big Questions
2. Computational Thinking—what is AI thinking?
3. GPT---what are the key AI ideas here?
4. Supervised learning
5. Scaling
6. Self-supervised learning
7. What’s next for AI?