Generating Reproducible Supplementary Materials With AI Help

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-research-AI-supplementary-materials-creators

1. Which of the following is NOT something AI can effectively generate when provided with the appropriate inputs?

   1. The raw experimental data itself that the analysis was performed on
   2. A data dictionary listing variables, types, units, and missing-value handling
   3. A reproducibility appendix matching a specific journal's format
   4. Code documentation with function signatures and expected behavior

2. A researcher uses AI to generate a reproducibility appendix, and the instructions sound completely plausible and professional. What should the researcher do before submitting?

   1. Submit immediately since the AI-generated content looks professional
   2. Replace the AI-generated content with manually written instructions
   3. Post the instructions online for community feedback before submission
   4. Have a colleague attempt to follow the instructions on a fresh machine

3. What is the primary purpose of a gap list in the AI-assisted supplementary materials workflow?

   1. To identify items that require author input rather than AI generation
   2. To track which data variables were excluded from the analysis
   3. To document which code functions were generated by AI versus written by humans
   4. To list journal formatting requirements for the submission

4. Why is testing reproducibility instructions on a fresh machine considered essential?

   1. Hidden dependencies and environment differences often cause failures that aren't apparent to the author
   2. Fresh machines have more processing power for complex analyses
   3. AI-generated instructions always work perfectly on first attempt
   4. Fresh machines are required by most academic journals

5. A researcher wants AI to generate code documentation for their analysis script. What inputs must they provide to the AI?

   1. Only the final publication PDF
   2. The analysis code itself and descriptions of what each function does
   3. A list of potential collaborators
   4. The journal's submission guidelines

6. Which component is typically included in a data dictionary for supplementary materials?

   1. Variable names, data types, measurement units, and how missing values are handled
   2. The complete raw dataset values
   3. Statistical significance values for all analyses
   4. A narrative description of the research findings

7. What responsibility remains with the author even when using AI to generate supplementary materials?

   1. Verifying the accuracy and correctness of all AI-generated content
   2. Obtaining copyright clearance for all referenced materials
   3. Hiring a professional editor for grammar review
   4. Writing the entire submission from scratch

8. What does a code repository README typically include?

   1. The full statistical analysis results
   2. The complete raw dataset in text format
   3. A list of all co-authors and their institutional affiliations
   4. Environment setup instructions and how to run the analysis

9. The lesson describes a common problem with AI-generated reproducibility instructions. What is this problem?

   1. They require expensive subscription services to access
   2. They are typically too technical for peer reviewers to understand
   3. They cannot be formatted to match journal requirements
   4. They often sound correct but contain functional errors when tested

10. What is the main advantage of using AI to generate supplementary materials drafts?

    1. AI can access and include proprietary datasets automatically
    2. AI eliminates the need for any human review of the materials
    3. AI allows authors to spend more time on verification rather than initial drafting
    4. AI guarantees that all journal requirements will be perfectly met

11. When generating a reproducibility appendix, why is it important to provide journal-specific requirements to the AI?

    1. Journals require AI-generated content to be labeled as such
    2. Journal requirements are standardized across all academic publications
    3. Different journals have different formatting and content requirements for reproducibility statements
    4. AI can only generate appendices for journals in its training data

12. What type of information should be documented in function signatures within code documentation?

    1. The email addresses of all users who have run the function
    2. Parameters, expected inputs, return values, and what the function does
    3. The exact time the function was last executed
    4. The entire source code of the function

13. What does 'missing-value handling' in a data dictionary specify?

    1. Which values in the dataset were excluded and the reason for exclusion
    2. The order in which data was originally collected
    3. How to handle reviewers who provide negative feedback
    4. How null, NA, or blank entries in the data should be interpreted and processed

14. Why might AI-generated documentation for an unseen dataset fail?

    1. The AI has no access to the actual data structure to understand variable relationships
    2. AI automatically corrects errors in unknown datasets
    3. AI refuses to work with datasets that haven't been preprocessed
    4. Datasets over 1GB cannot be processed by AI systems

15. In the context of this lesson, what is the 'bottleneck' that supplementary materials represent?

    1. The slow internet connection used to submit materials
    2. The cost of preparing supplementary materials
    3. The point in the submission process where time is often rushed and quality suffers
    4. The journal's review process for supplementary files