Skip to content
Idaho AI Classroom
All lessons
AI ImpactImpact on Society9-12.AII.IS.2

Idaho Work, AI Tools

11–1250 minutes90 minutes (extend seminar and writing)

Standard quoted exactly

Analyze how workers in different careers use AI to solve problems.

Student-friendly learning targets

  • I can analyze how a specific Idaho worker uses AI to solve a defined problem, not merely list apps.
  • I can compare at least two careers on what the human still must judge after the tool runs.
  • I can plan or conduct an ethical interview or profile study that does not collect extra personal data.

Essential questions

  1. What problem is the worker actually trying to solve, and where does the tool sit in that problem?
  2. What does a nurse, a grower, a technician, a journalist or teacher, and a Micron-area manufacturing or software worker still have to know that the model does not?
  3. How should a student talk to a working adult about AI without turning the visit into a product demo?

Objectives

  1. Study five career profiles: clinic/nursing, farm/agronomy, mechanic/technician, journalism or teaching, and software or semiconductor manufacturing (Micron-area).
  2. For each, name a problem, a tool or method, a human judgment that remains, and a risk if the tool is wrong.
  3. Conduct a structured interview or a packet-based profile study using the same questions, not a free-form chat about favorite apps.
  4. Compare two careers on data sensitivity (patient records, yield data, student work, factory telemetry).
  5. Produce a one-page analysis of a single worker’s problem-solving loop: Human aim → data → tool → check → decision.

Key vocabulary

Problem of practice
The concrete job to be done (triage a symptom pattern, time an irrigation set, diagnose a fault code, verify a quote, catch a yield drift), not the name of a software product.
Decision support
A tool that ranks or drafts options while a qualified human remains responsible for the call.
Domain knowledge
The field-specific understanding that lets a worker notice when an output is fluent and wrong.
Telemetry
Instrument data from machines, lines, or fields (vibration, moisture, temperature, throughput) used to detect problems.
Clinical judgment
A licensed professional’s responsibility for patient decisions, which a model cannot hold.
Profile study
A structured analysis of a worker’s process using documents and prepared notes when a live interview is not possible.
Interview protocol
A short list of questions asked the same way to different people so comparisons are fair and privacy is respected.

Teacher background

The standard is about workers solving problems, not about a parade of brand names. Idaho gives you five nearby worlds if you look: a nurse or medical assistant in a St. Luke’s, St. Alphonsus, or rural clinic; an agronomist or grower on potatoes, sugar beets, dairy, or cattle; a diesel or industrial technician in the Treasure Valley; a reporter or a classroom teacher; and a process, yield, or software role tied to Micron and its suppliers. Students should leave able to narrate a loop: the human has a problem, some data exists, a tool proposes a pattern or a draft, a qualified person checks, a decision is made. If the check disappears, the analysis has failed. Live interviews are gold when a guest can come, or when a student already works with a family adult who consents. Many rural and small-town classes will need printed profiles and a recorded backup; that is a profile study, not a lesser task, if the questions are the same. Do not let the hour become a list of apps. Do not ask guests for patient names, student names, unpublished yield numbers, or factory secrets. FERPA and health-privacy rules travel with the clinic and the school profiles. Offline fallback: five one-page profiles written by the teacher from public career materials, plus an interview protocol students can run later with consent.

Materials and prep

Materials

  • Five printed profiles (nurse/clinic, farmer/agronomist, mechanic/technician, journalist or teacher, Micron-area software or manufacturing). Each names a problem, a tool type, a human check, and a risk.
  • Interview protocol (same eight questions for every career). Consent line at the top.
  • Problem-solving loop poster: Aim → Data → Tool → Check → Decision.
  • Comparison grid for two careers: data sensitivity, cost of error, who is licensed to decide.
  • Offline fallback: recorded two-minute clips if a guest exists; otherwise packets only. No student is required to have a parent in a listed industry.

Before class

  • Invite a guest if possible and record a backup for activity buses and bandwidth failures.
  • Rewrite profiles so they are realistic but not identifiable as a particular neighbor.
  • Send the protocol home for optional adult interviews with a consent note: no patient, student, or proprietary data.
  • Block plan: two interviews or two deep profile studies, then a comparison essay.

Instructional sequence

Name the problem, not the app

5 min
  1. Students list any AI they have heard a worker use. Cross out product names. Circle the problem underneath (catch a bad part, draft a parent email, time a pump).
  2. If a student cannot name a problem, that is data: they have seen marketing, not work.
  3. Show the five Idaho worlds on the board. Tell them every student will leave with one full loop, not five logos.
  4. Read the standard. Underline workers and solve problems.

Five worlds, one loop

10 min
  1. Walk one clinic example: a triage note drafted from structured fields, checked by a nurse, never a diagnosis from a public chatbot, patient identifiers stripped.
  2. Walk one agronomy example: satellite or soil telemetry suggesting a variable-rate pass; the grower still knows the field’s low spot that the model treats as noise.
  3. Walk one technician example: a fault-code suggestion list; the mechanic still tests the cheap failure first and does not replace a part because a screen said so.
  4. Walk one journalist/teacher example: transcription and a first-pass summary; quotes still get verified; student essays still get a human reader for thinking, not just grammar.
  5. Walk one Micron-area example: yield or defect pattern detection on a line; a process engineer still owns whether to stop the tool. No proprietary numbers in class.

Profile the nurse together

10 min
  1. Read the clinic profile aloud. Fill Aim → Data → Tool → Check → Decision on the board.
  2. Ask what data must never enter a consumer model (names, dates of birth, conditions). Connect to FERPA-like duties and health privacy.
  3. Ask what happens if the check is skipped on a night shift. That sentence is the analysis.
  4. Model citing the profile as a source, not I think nurses use ChatGPT.

Study one profile; compare with a neighbor

12 min
  1. Each student takes one remaining profile and writes a one-page loop analysis.
  2. Pair with someone who had a different career. Complete the comparison grid on data sensitivity and cost of error.
  3. If a consented guest is present, two students ask protocol questions only; others take notes as a profile study.
  4. Ban app lists. If a sentence names a brand without a problem, rewrite it.

Real-world examples

  • A rural clinic using speech-to-text for notes, with a nurse correcting terms the model misses, never pasting a chart into a public chatbot.
  • A Magic Valley dairy or potato operation using imagery and soil data to decide a pass, with an agronomist walking the field before spending money.
  • A Nampa or Idaho Falls technician using a diagnostic suggestion list, then confirming with a test, not a guess.
  • A reporter at a statehouse or school-board meeting using a transcript, then calling to confirm a quote.
  • A process engineer in the Boise-Meridian semiconductor corridor watching defect clusters, with a human stop-the-line authority.

Hands-on activity

Interview dry run

8 min
  1. Pairs practice the protocol in role-play: one is the worker from their profile, one is the student. Three minutes each.
  2. Listeners mark when a question asked for PII or a trade secret and rewrite it.
  3. Class collects three excellent questions on the board (problem, check, failure case).
  4. Assign the optional homework interview only with written adult consent.

Discussion questions

  1. What did every career still need a human to do after the tool ran?
  2. Which job’s error costs a life, a crop year, a crashed engine, a false headline, or a scrapped wafer — and how does that change the check?
  3. Why is listing apps a weaker analysis than describing a problem of practice?
  4. How should a student intern refuse a request to paste customer or patient data into a public model?
  5. What Idaho career not on the five-card set would you add, and what problem would you study?

Differentiation

Support

  • Provide a one-page loop template with sentence starters.
  • Allow a student to stay with the clinic profile after guided practice instead of switching.

Challenge

  • Add a sixth Idaho world (INL, tribal enterprise, logistics, energy) with sources cited.
  • Write the comparison as a memo to a counselor explaining which pathway still requires deep domain knowledge.

Multilingual learners

  • Profiles should include at least one worker who uses Spanish or another language on the job; analyze how translation tools help and fail (link EP.2).
  • Interview protocol available with simplified English; students may ask questions in the adult’s preferred language if both consent.

IEP / 504

  • Oral loop analysis is acceptable. Role-play may be written instead of spoken if needed.
  • Do not require a home interview; the packet study meets the standard.

Assessment

Formative

  • Warm-up problem-not-app rewrite.
  • Comparison grid completeness.

Summative

  • One-page loop analysis of a single career, scored on problem, tool role, human check, and risk.
  • Block extension: two careers compared in a 400-word seminar paper, plus notes from a live or recorded interview.

Success criteria

  • Names a problem of practice, not only a product.
  • Includes a human check and a failure cost.
  • Uses Idaho-relevant work (clinic, field, shop, classroom or newsroom, semiconductor/software) with respectful privacy limits.

Responsible use, ethics, and privacy

Responsible use

No patient, student, or proprietary factory data in notes or tools. Guests may refuse any question. Students do not install workplace software on personal phones for this assignment. Offline profiles fulfill the standard. Recorded backups beat failed video links.

Ethics

Workers are experts in their problems. Students analyze how tools sit inside that expertise; they do not instruct a nurse or a grower. Job displacement may come up; treat it as an evidence question (what task, what check, what hiring pattern), not as destiny or as a pep talk.

Privacy

Clinic stories follow health-privacy rules. School stories follow FERPA. Factory and farm numbers may be confidential. Interview notes store no extra identifiers (children’s names, addresses, account logins). Optional interviews need documented consent and may be declined without grade penalty; the packet study remains available.

Reflection

  1. Which human check would you want if you were the patient, the grower, or the reader?
  2. What question in the protocol produced the most useful answer in role-play?
  3. Where were you tempted to list an app instead of a problem?

Homework

Optional: run the protocol with a consented adult (fifteen minutes, no sensitive data). Required if no interview: annotate a second printed profile with the loop and one question you would still ask. Do not look up employees on social media to complete this.

Closing

Read one Aim → Decision loop aloud. Collect analyses. Tomorrow traces how bias enters long before a worker sees an output — at every stage of working with data.

Extensions and cross-curricular links

Go further

  • 90-minute block: guest or recording plus a comparison essay (seminar and writing).
  • Coordinate with CTE, FFA, HOSA, or a counseling career unit so the profile is not an island.
  • Students who complete a consented interview attach notes and a thank-you; no recording without extra consent.
  • Map tools to the three tests from IS.1: how would this worker check bias, accuracy, and harm?
Health science / HOSA
Clinical judgment, documentation, and why public chatbots are the wrong place for chart fragments.
Agriculture / FFA
Agronomic decision support, field-truthing, and data that is a business secret.
Automotive / industrial CTE
Fault codes as suggestions, tests as evidence.
English / journalism
Verification of quotes and the difference between a transcript and a story.
Engineering / manufacturing
Yield, defects, and stop-the-line authority in semiconductor work.