Week 1 — How AI Actually Works

Learning objectives

Core concepts

AI systems are pattern-matching engines trained on huge datasets. They do not “know” things the way a person does — they produce statistically likely outputs based on what they’ve seen before. This single idea explains most of what comes later in the course: why AI hallucinates, why it reflects bias, and why it can sound confident while being wrong.

Discussion prompts

  1. Before this week, what did each of you assume AI was actually doing when you asked it a question? How close was that to the reality?
  2. Where have you noticed AI showing up in apps you already use? Make a list together.
  3. If an AI is just predicting the next likely word, what kinds of questions is it probably good at — and what kinds is it probably bad at?
  4. Teen to parent: what’s one thing you wish adults understood about how your generation uses AI? Parent to teen: what’s one thing you wish you understood better?

At-home activity: “Same question, three tools”

Pick one open-ended question you both genuinely care about (e.g. “What should we cook for dinner this week on a $60 budget?” or “What were the causes of WWI?”). Ask the same question in three different tools — for example ChatGPT, Google Gemini, and a regular Google search. Compare the answers side by side: - Where do they agree? - Where do they disagree? - Which one cites sources? Are the sources real? - Which felt most confident? Was confidence the same as correctness?

Write down one sentence each about what surprised you.

Parent resource list

Reflection

Write one sentence: “Before this week I thought AI was ___; now I think it’s ___.”