AI Fluency: Foundations

Accurate intuitions about what you're working with — no hype, no doom, just functional understanding.

• The AI landscape: AI ⊃ ML ⊃ Deep Learning ⊃ LLMs/Diffusion
• Discriminative vs. generative AI
• How LLMs work (intuition only — no math)
• How diffusion models work
• Why AI behaves the way it does: tokens, context, training cutoffs
• Tour of the AI ecosystem
• When to use which AI tool
• Three modes of interaction: Automation, Augmentation, Agency
• Project: Map the AI Landscape

Three frameworks side by side so students can pick the one that clicks.

• Why frameworks matter (and when they don't)
• The 4D Framework (Anthropic): Delegation, Description, Discernment, Diligence
• TCREI (Google): Task, Context, References, Evaluate, Iterate
• The Pattern Library approach (Vanderbilt)
• Comparing the three: when each shines
• Building your own hybrid workflow
• Project: Apply a Framework

When not to use AI is just as important as when to.

• Goal awareness: what does "done" look like?
• Task decomposition: AI vs. human vs. collaborative
• The feasibility × value matrix
• Tasks not jobs: AI's actual unit of impact
• Platform selection: model strengths and limitations
• Cost considerations: tokens, subscriptions, time, attention
• When NOT to use AI
• Project: Delegation Audit

Depth, clarity, and a memorable structure for the most-saturated topic in AI education.

• Anatomy of a good prompt
• Product description: describing the output you want
• Process description: dialogic, iterative prompting
• Performance description: defining future AI behavior
• Zero-shot prompting
• Few-shot prompting (examples)
• Chain-of-thought prompting
• Role and persona prompting
• Structured output (JSON, tables, schemas)
• Prompt chaining and decomposition
• Multimodal prompting (images, voice, files)
• Iteration: when to refine vs. start over
• Project: Build a Prompt Library

A full module on output evaluation — most courses bury this in 1–3 lessons.

• What you're really evaluating: quality vs. accuracy vs. usefulness
• Spotting hallucinations: cognitive heuristics
• The Fact-Check List Pattern
• Verifying sources and claims
• Spotting bias in AI outputs
• The Description-Discernment loop
• Process discernment: is this collaboration even working?
• When AI confidently lies: failure modes you'll see often
• Project: Output Evaluation Rubric

• Creation diligence: stakeholder impact, fairness, edge cases
• Bias as a societal concern
• Copyright, IP, and the training data debate
• Privacy: what you should never paste into AI
• AI security at the user level: prompt injection, data leakage, jailbreaks
• Disclosure norms: when and how to declare AI involvement
• Deployment diligence: verifying before shipping
• Job impact and the tasks-not-jobs framing
• Building your personal AI policy
• Project: Your Personal AI Policy

Brief conceptual exposure to advanced topics — recognition, not deep dives.

• What are AI agents? (concept, not implementation)
• The four agent design patterns: reflection, tool use, planning, multi-agent
• RAG explained: giving AI a custom library to read from
• Memory and context management
• When prompting isn't enough: signs you need agents, RAG, or fine-tuning
• Project: Design an Agent Workflow

Pick a real project. Use everything from M1–M7. Ship something. Document the process.

• Capstone Project