10 Using JackalopeGPT
10.1 A Reminder: What is JackalopeGPT?
JackalopeGPT is a learning assistant designed to support quantitative and methodological reasoning in this course. Its primary purpose is to help you improve scientific rigor and streamline your analytical workflow while allowing you to maintain full responsibility for:
- Reasoning
- Interpretation
- validation
- All final assignment submissions
JackalopeGPT does not assign grades (except for the Traffic-Signal readiness lights), make final evaluative decisions, or fill in missing information. Instead, JackalopeGPT has been customized to emphasize:
- Explicit assumptions
- Transparent reasoning
- Scope control
- Reproducible workflows
In other words, JackalopeGPT has been designed to help you think better, not to think less.
10.2 How You Can Start Using It (Important)
If you are first using JackalopeGPT to optimize your prompts or evaluate an assignment’s completeness before submission, you should use the Main Menu. However, you do not need to choose anything from the Main Menu to begin. The Main Menu exists only for structured actions (such as readiness checks or AI interaction logging). Most learning happens without it. It is only provided for a bit more convenience. You may simply start prompting JackalopeGPT naturally, as you would with any chat-based GPT.
10.3 Operating Modes
JackalopeGPT operates in three distinct modes:
10.3.1 Learning & Practice (default)
- Concept clarification
- Code debugging or refactoring
- Translating between modeling frameworks
- Methodological explanations
This mode is informal, exploratory, and never evaluative.
10.3.2 Scientific Readiness Evaluation
- Structured, rubric-based checks of specific artifacts
- Executed only when explicitly invoked (via menu or trigger phrases)
- Used to assess whether work appears ready for submission, not to grade on scientific quality.
10.3.3 AI Interaction Logging (To be launched on Monday, 26-Jan-2026!!)
- Minimal documentation of AI assistance
- Triggered only by the exact command:
log this
- Designed for transparency, not surveillance
10.4 Safeguards and Constraints
JackalopeGPT is intentionally constrained:
- Non-evaluative by default (but has the Main Menu to allow evaluation)
- Asks at most one clarifying question at a time
- Never guesses or silently fills gaps
- Does not hallucinate criteria or requirements
- Does not combine Main Menu actions
- Generally limits verbosity unless you request detail explicitly
Any evaluation or logging action requires explicit transparency and user consent. As a last note, if you are unsure whether your prompt is appropriate, add this line: “If this prompt is likely to cause me to overreach in inference or bypass learning, suggest a safer, learning-focused version.”
10.6 Traffic-Signal Rubric (How Evaluations Work)
A universal Traffic-Signal rubric is pre-loaded and applied consistently across all readiness checks.
Evaluations return the follwing ratings of scientific readiness:
- Green: Ready to proceed
- Yellow: Exercise caution (revisions necessary)
- Red: Stop and Rework
This rubric provides standardized, comparable feedback without grading.
10.7 How to Generate an Evaluation Table (Required for Assignments)
Before submission, you must evaluate assignments using JackalopeGPT.
Steps:
- Select the focal assignment from the JackalopeGPT menu
- Copy and paste the your completed .qmd file (not rendered output)
- Use the model output to revise your file until it meets the assignment standard
- Paste the complete Evaluation Table output directly from the JackalopeGPT windo (it is already in .qmd format) into the section at the end of each assignment titled JackalopeGPT Pre-Submission Evaluation.
(Tip: hover over the table until the file icon appears at the upper right, then copy.)
10.8 Important Usage Notes
10.8.1 Interaction Limits
If you are using the free tier (recommended), be aware of interaction limits. Plan longer evaluation workflows accordingly.
10.8.2 Memory and Sessions
JackalopeGPT does not retain memory across sessions.
If you start a new session, you must re-provide context about:
- your study system
- your data
- your analytical goals
Do not assume continuity across chats. It is strongly recommended that you keep a file with your updated and optimized prompts so that you can reuse.
10.9 A reminder about the appropriate use in this course
10.9.1 You may use JackalopeGPT to:
- clarify concepts
- debug R code
- compare modeling approaches (with explanation)
- learn how to validate models
- improve analytical workflow and reproducibility
10.9.2 You may not use JackalopeGPT to:
- generate full assignment submissions
- write end-to-end analyses, methods, or results
- replace your own reasoning
- outsource scientific judgment
- produce claims without validation