AI Session Card — Template

AI Session Card — Template#

Copy this template into ai_logs/lab7_session_NN.md (one file per session) and fill in the bracketed sections. The session cards together form a portfolio artifact you will reference in your final project’s reproducibility statement.

session_id: lab7_session_NN
date: YYYY-MM-DD
tool: [ChatGPT-4 / Claude 3 / Copilot / other — include version if known]
duration_minutes: NN
context_block_used: [yes / no — paste the block you used]

What I asked the AI to do#

[One or two sentences describing the task. Be specific — not “help me with refraction” but “derive the intercept-time formula for a flat 2-layer refractor, step by step.”]

My exact opening prompt#

[paste verbatim — including any context block]

What the AI returned (summary)#

[3–5 bullets summarising the response. Do NOT paste the full response — instead, capture the gist.]

What I cross-checked, and against what#

[Name the specific page, equation, or notebook cell you verified the response against. Examples:

  • “Lowrie & Fichtner Ch. 4 §4.2.2, equation 4.34”

  • “Lab 3 notebook, cell labelled # Step 5 — fit V2”

  • “ESS 314 lecture 06_seismic_refraction_I, §3 Mathematical Framework”]

What the AI got right#

[Bullet list.]

What the AI got wrong, or what it missed#

[Bullet list. If nothing, write “Nothing detected at this depth of checking — flag for re-verification later.”]

What I would do differently next time#

[One sentence on prompt strategy, e.g., “I would supply the two-layer geometry as a sketch description first, before asking for the derivation.”]

Citation block for my final report#

[A short, paste-ready citation, e.g.: “ChatGPT-4 (OpenAI). Conversation on derivation of refraction intercept time. Personal communication, 5 May 2026. Used to scaffold §3.1 of methods section; all equations independently re-derived from Lowrie & Fichtner Ch. 4.”]


---

## Why these fields?

- **`tool` + `date`**: Required by the course AI use policy.
- **Context block**: Forces you to use Pattern 1 consistently.
- **Exact opening prompt**: Lets you (and a grader) see whether your
  prompt was well-specified or vague.
- **Cross-check against**: This is the *Stage 2 → Stage 3* discipline.
  An AI session with no cross-check is *not* a learning session.
- **Got wrong / missed**: Building the catalogue of AI failure modes
  is itself a learning artifact.
- **Citation block**: This is what goes in your final report's
  acknowledgements or methods section, per the ESS 314 syllabus.