CSCA 5112: Introduction to Generative AI

Duration: 3 hours

In this first week, you will meet generative AI the same way you might meet a curious stranger on a bus—through open, playful conversation and low-pressure experimentation. You will jump straight into using text, image, and audio tools to explore whatever is on your mind, then step back to learn how language models actually generate responses, what prompts are, and why context matters. With a clearer understanding of how these tools work and where they fit in the broader story of AI, you will return to your experiments to refine your prompting and improve your results. The week emphasizes curiosity, iteration, and intuition-building, and closes with an AI-guided reflection to help you clarify what you want to gain from the course and how generative AI can support your goals.

Duration: 4.5 hours

In Week 2, you will discover that not all AI tools are created equal and that choosing the right one can completely change what’s possible. You will compare how different tools handle the same tasks across text, images, audio, and code, and build intuition for how underlying model types like transformers, diffusion models, GANs, and VAEs shape what a tool can generate and why that matters. Without getting buried in technical complexity, you will explore big ideas like embeddings and retrieval-augmented generation (RAG) in ways that connect directly to real use. Using NotebookLM, you will generate a personalized podcast from course materials to experience how RAG reshapes outputs based on your needs, then reflect on its limits and where human judgment still plays a critical role. By the end of the week, you will be thinking more strategically about how to match the right model to the right job.

Duration: 1.5 hours

In Week 3, you will level up how you work with generative AI by learning how to guide it with intention through prompt engineering and your first steps into context engineering. You will explore how models use attention, what a context window is, and how tokenization shapes what the model actually “sees,” helping you understand why small changes in structure can lead to big changes in output. You will move beyond simple prompt tips into designing prompts with clarity, roles, examples, and sequencing, while also learning how conversation history influences results. Through a hands-on prompt-and-refine loop, you will iteratively strengthen your inputs, experiment with managing or resetting context, and sharpen your strategy in a gamified “Prompt Jeopardy” challenge. By the end of the week, you will not only write better prompts you will understand why they work and how to shape outcomes more consistently and powerfully.

Duration: 3 hours

In Week 4, you will zoom out from building skills with generative AI to examine its limits, risks, and ethical implications with a more critical lens. Using the “Three R’s”: Responsibility, Red Flags, and Retrieval-Augmented Generation (RAG). You will learn how to recognize hallucinations, bias, training cutoffs, and context limits, and how techniques like grounding models with external data can improve accuracy without removing the need for human judgment. You will revisit the “stranger on the bus” metaphor in a deeper way, exploring what it really means to interact with a system that sounds confident but is also learning from collective human behavior. Through case studies, hands-on experiments, and discussion, you will begin to see GenAI not as a magic box, but as a powerful tool shaped by design choices, tradeoffs, and human responsibility marking a clear shift toward more intentional, ethical use.

Duration: 3 hours

Final Exam Format: "mini-capstone" project (in-course)

This module contains materials for the final exam. If you've upgraded to the for-credit version of this course, please make sure you review the additional for-credit materials in the Introductory module and anywhere else they may be found.