CS-215 AI Literacy · Week 1 · Meeting 1

Evidence Card: Task + Minimum Context

Use this on paper or annotate the PDF. It is ungraded and not submitted. The web page does not save or transmit responses.

Name or initials (optional)Date

Privacy boundary: use only the fictional course scenario. Do not record diagnoses, medications, addresses, credentials, student records, or another person’s private information.

1. Task map sentenceSlide M1.3

AI could help with ___, but a person should remain responsible for ___ because ___.

2. Minimum-context promptSlide M1.5

Draft the missed-class email using the minimum relevant fictional context. Do not invent an entitlement or deadline.

3. Omission rationalePrivacy check

What did you intentionally leave out, and why was it unnecessary?

CS-215 AI Literacy · Week 1 · Meeting 1

Prediction, Tokens, Context, and Exit Model

Use plain language. Technical vocabulary matters only when it helps you explain the idea accurately.

4. Prediction is not truthSlides M1.7–M1.8

Explain why a fluent continuation can still require verification.

5. Token versus toolSlide M1.9

A token is…

A tool is…

6. Four-part context checkSlide M1.10

AVAILABLE

What information can this product actually access now?

RELEVANT

Which supplied information directly helps this task?

AUTHORITATIVE

Which source is allowed to settle the claim?

CURRENT

How will you know the source still applies?

7. Meeting 1 two-sentence mental modelExit

Sentence 1: what an LLM does. Sentence 2: why useful language may still need evidence.

CS-215 AI Literacy · Week 1 · Meeting 2

Post-Training + Claim/Evidence Ledger

Evidence scope: classify each claim only against the course-authored fictional packet shown in class, checked 30 August 2026. These labels do not make claims about a real professor, policy, or current university rule.

8. Complete the post-training sentenceSlide M2.3

Post-training can change ___, but it does not automatically ___.

Evidence-label key

SUPPORTEDThe supplied packet directly supports the claim.

CONTRADICTEDThe supplied packet conflicts with the claim.

UNSUPPORTEDA present assertion is made, but the packet does not support it.

NOT YET KNOWABLEThe answer depends on a future or authorized decision that has not happened.

9. Four-row claim/evidence ledgerSlide M2.5

Claim	Evidence	Label	Correction or action
Proposal due Friday at 5:00 p.m.
Every sick student automatically receives a makeup.
Professor Lee guarantees a reply within 24 hours.
Professor Lee will approve this request.

10. Corrected summaryUse only supported strength

CS-215 AI Literacy · Week 1 · Meeting 2

Verification Priority + System Authority

11. Verification-priority rationaleConsequence · urgency/action · evidence · reversibility

Which claim should receive attention first? More than one answer can be defensible; use the four factors in your reason.

12. Label the system componentsSlide M2.7

Scenario: the assistant retrieves the approved syllabus, cites it, drafts an email, and waits for the student.

□ LLM

□ Chat interface

□ Tool

□ Agent loop

□ Human approval

Other: ________

13. Permission-boundary sentenceSlide M2.8

The system may ___, but it must not ___ without ___.

14. One Proceed / Ask / Block caseSlide M2.9

Case	Decision + visible fact that controls it
Round ___

15. Week 1 exit revisionTwo sentences

What idea did you revise? What rule will you use when AI gives you a polished answer?

Optional weekend assignment: □ No homework □ Quick Review: 8-minute video + one question □ Extended Review: selected video segments + six responses. Neither route is graded. If you complete one, submit its PDF in eCampus for formative feedback; choosing no homework carries no penalty.