Artificial Intelligence @ Johnson

Section NBA 6925, Spring 2026 (second half)
Instructor Emaad Manzoor (emaadmanzoor@cornell.edu)
Meeting Location Breazzano Center, Room TBA
Meeting Time 11:40AM-12:55PM, Tuesday & Thursday
Canvas Site TBA (Syllabus)
Credits 1.5

This course is a continuation of NBA 6921 for builder-focused students, and covers advanced topics in modern artificial intelligence. It includes a course project. Passing NBA 6921 is a prerequisite.

Graded Work

The final grade will be a function of the following components.

  1. Homework 3 (20%): See PDF.
  2. Homework 4 (20%): See PDF.
  3. Group Project (30%): See project requirements.
  4. Exam (15%): Open laptop, ChatGPT etc. allowed.
  5. Attendance (10%): Attendance is mandatory and will be recorded.
  6. Course Evaluation (5%): See PDF.

The average grade in this section will be 3.50 as per Johnson School policy. Grades are relative and will be distributed as follows:

  1. Top 5%: A
  2. Next 50%: A- (modal grade)
  3. Next 25%: B+
  4. Next 10%: B
  5. Next 10%: B- and lower

Schedule

Date Topic Notes
Mar 12 NBA 6921 Course Recap HW3 out
Mar 17, 19 Post-training, Alignment, and Safety
    Preference Optimization
    Prompt Injection and Jailbreaking		 Project proposal due
Mar 24, 26 Project Hackathon
    Groups work with me in my office
    to implement their project demo
    Come with your idea and data!		  
  Spring Break  
Apr 7 Multimodal and Embodied AI
    Diffusion Models
    VLMs, VLAs, and World Models
    Embodied Agents		  
Apr 9 Mechanistic Interpretability
    Probes and Circuits
    Model Editing
    Hands-on: Transformer Circuits		 HW3 due
HW4 out
Apr 14, 16 Augmented Intelligence
    Diagnosing Decisions with AI
    Designing Human-AI Collabs
    Case Study: Pinterest		  
Apr 21 Practice Exam  
Apr 23 Exam
    Open laptop, ChatGPT etc. allowed
    No human-human collaboration		  
Apr 28 Recommendation Machines
    Factorizing Interaction Matrices
    The Cold-start Problem
    Missingness and Feedback Loops
    Case Study: Spotify		 HW4 due
Apr 30 Causal Artificial Intelligence
    Causal Inference: The “What If” Problem
    Heterogeneous Treatment Effects
    Personalized Treatment Policies		  
May 5, 7 In-Class Project Presentations Slides due