Suggested sequence
Pacing guide
The standards do not prescribe curriculum. This sequence is one local option: Fluency for 9–10, Impact for 11–12, and Technical Aptitude as a concurrent CS/engineering strand. Adjust for block schedules, trimester calendars, and rural connectivity.
01 · Semester 1 · Weeks 1–4
9–10 · 4 weeks
What AI is — and is not
Appropriate use, model families, physical vs. simulated worlds
Open with judgment, not tools. Students should leave this unit able to name when a calculator, a search, a human, or an AI is the right move.
02 · Semester 1 · Weeks 5–9
9–10 · 5 weeks
Think with it, not for it
Cognitive offloading, critique, hallucination, sensory grounding
- What We Lose When We Hand It Over
- A Second Reader, Not a Ghostwriter
- Catching Hallucinations
- Bias, Limits, and Blind Spots
- Senses vs. Statistics
Pair every generation task with a verification task. Academic-integrity conversations belong here, before high-stakes writing season.
03 · Semester 1 · Weeks 10–14
9–10 · 5 weeks
Agents, code, and communication
Agents, simple code generation, interpreting output for an audience
- What an Agent Actually Does
- Run It, Don't Worship It
- Build a Narrow Agent
- Translate the Machine
- Idaho Futures, AI Skills
Keep code projects tiny and runnable. Career connections should name Idaho pathways: agriculture, energy, health, trades, public service, and software.
04 · Semester 1 · Weeks 15–18
9–10 (bridge to 11–12) · 4 weeks
Human, machine, and responsibility
Debate, training-data bias, and a fluency capstone
Use the debate as a performance assessment. The same prompt returns in 11–12 at greater depth (AII.IS.4).
05 · Semester 2 · Weeks 1–5
11–12 · 5 weeks
GenAI, ethics, and private lives
Ethics of commonplace GenAI, accessibility, data collection, unauthorized training
- After the Flood: GenAI in Every Subject
- Who Gets a Better Door
- The Cost of a Billion Examples
- Used Without Asking
Invite a counselor or librarian for the FERPA/COPPA conversation. Do not ask students to upload personal writing into unapproved tools.
06 · Semester 2 · Weeks 6–10
11–12 · 5 weeks
Work, harm, and civic choice
Cybersecurity, output evaluation, careers, data-stage bias, regulation
- Defense, Offense, and Dual Use
- Three Tests: Bias, Accuracy, Harm
- Idaho Work, AI Tools
- Bias Enters at Every Stage
- Who Decides, Who Pays
Idaho guest speakers (Micron, agriculture co-ops, hospitals, county IT, tribal enterprises) make IS.2 concrete. Record a backup video for rural bandwidth days.
07 · Semester 2 · Weeks 11–13
11–12 · 3 weeks
The hard debate, revisited
Consciousness, creativity, ethics, and human responsibility at depth
Require evidence from prior units. Students should cite at least one technical limit and one civic consequence.
08 · CS / Engineering strand · ongoing
9–12 CS · 6–8 weeks
Data before models
Cleaning, validation, acquisition, train/dev/test, bias in examples
- Make the Text Behave
- When a Number Is a Word
- Collect, Split, Then Train
- From Web Page to Table
- Better Examples, Fairer Model
- Critique the Corpus
- The Lifecycle, Methods View
Spreadsheet-first. Python notebooks are an extension, not a gate. Public datasets only; no student PII.
09 · CS / Engineering strand · ongoing
9–12 CS · 6–8 weeks
Methods, metrics, and safeguards
Model families, XOR vs. language, prebuilt models, agents, metrics, human-in-the-loop
- Pick the Model That Fits
- Call a Model, Don't Train One
- Drop In an Agent
- Data Writes the Function
- Rules, Data, or Both
- Three Families, Three Jobs
- What This Model Cannot Do
- Human in the Loop, On Purpose
- Accuracy Is a Trap
Teach AIM.9 (metrics) before any demo that quotes “99% accuracy.” Fraud and wildfire examples make class imbalance visceral.