To compare the requested lists
When parties file separate requested jury instructions that simply list instruction numbers, one can enlist ChatGPT to review both filings and highlight any differences.
Prompt:
List all differences between the requested instructions in these two lists.
Do not include source links, URLs, or inline citations in the response.
Return the final work as a downloadable .docx file and provide the download link.
ChatGPT will create an informative, downloadable Word file that looks like this. In this example, ChatGPT took 5 minutes in extended-thinking mode. This example is of a demo Michigan criminal case.
To combine the requested lists into one
Now, let’s say you want to combine both lists into a single Word document.
Next prompt:
Now, combine both lists into one comprehensive list, displayed in numerical order.
Do not include source links, URLs, or inline citations in the response.
Return the final work as a downloadable .docx file and provide the download link.
ChatGPT will create a helpful downloadable Word file that looks like this. In this example, ChatGPT took 6 minutes in extended-thinking mode.
To create full jury instructions from one list
Ready for the next step and have ChatGPT take your list and pull in the full-instruction content from the current version of the Michigan Criminal Jury Instructions (and bold those sections that will still require customization for the case)? You can do that! (You can also do this using the civil jury instructions after you make appropriate adjustments to the prompt.)
Create a sample Word/.docx file of a couple of instructions to show how you want your output formatted in Word. Here, I call my file “JI Format Template.docx”. Upload that to ChatGPT. And attach your jury instruction list that you want prepared.
Next prompt:
I’m attaching:
1) “JI Format Template.docx” (this is the FORMAT TEMPLATE). Match its styles, spacing, indentation, margins, headers/footers, and pagination exactly.
2) A file containing the list of M Crim JI numbers to include, in the exact order listed. Do not reorder, omit, or add anything.
For the full and current text of each listed Michigan Model Criminal Jury Instruction, use the official PDF here:
https://www.courts.michigan.gov/48d3b7/siteassets/rules-instructions-administrative-orders/jury-instructions/criminal/current/criminal-jury-instructions.pdfTASK:
Create a NEW .docx that matches the template formatting exactly.For EACH M Crim JI number in my list, create a section that starts on a NEW PAGE (insert a page break before each new instruction section) and includes, in this order:
1) Line 1: “M Crim JI X.XX” (use the same style as the template’s instruction heading line)
2) Line 2: [PASTE THE OFFICIAL INSTRUCTION TITLE ONLY] — DO NOT include the word “TITLE”, and do not use “TITLE:” or any label.
3) Then paste the FULL instruction text verbatim from the official PDF (not a placeholder). Preserve the instruction’s paragraph breaks, numbering/lettering, and internal indentation. Apply the template’s body style so I can edit/paste without reformatting. When pasting the full instruction text, apply the template’s Body/Text style and confirm it retains the template’s 1.5 line spacing (Paragraph → Indents and Spacing → Line spacing). If the pasted content overrides spacing, reapply the Body style and then force the paragraph line spacing to match the template (1.5)
4) Bracketed text formatting (mandatory): In the instruction text only (not the “M Crim JI X.XX” line and not the title line), bold every character that appears between square brackets, including the brackets themselves.
- Example: … [Choose one: …] … → … **[Choose one: …]** …
- Apply this even if the bracketed text spans multiple words, punctuation, or line breaks.
- Do not bold parentheses ( ), curly braces { }, or quotation marks—square brackets only.
- Do not change any wording, spacing, numbering, indentation, or line breaks; only apply bold to the bracketed content.
5) Verification: After generating the .docx, check a random sample of paragraphs in multiple sections to confirm 1.5 line spacing is in effect (same as the template), not reset to single.
OUTPUT REQUIREMENTS:
– Do not include any placeholder text.
– Do not include any commentary, notes, or extra pages beyond the instruction sections.
– Provide the final document as a downloadable .docx.
– Filename must be: “[CASE NUMBER]_Proposed JI-Party name.docx”
CASE NUMBER: ENTER CASE NUMBER HERE
ChatGPT will create a helpful downloadable Word file that looks like this. In this example, ChatGPT took 22.5 minutes in extended-thinking mode.
To compare filed jury instructions against the current model jury instructions
Other times, a party will file full, completed instructions, but you need to compare them against the current model jury instructions for accuracy and completeness. Why? Because counsel might have simply “recycled” instructions from an earlier trial, yet some model instructions may have since been updated—and counsel did not verify currentness before filing.
This new prompt can help in that endeavor.
I am providing two sources:
Document 1: the complete jury instructions filed by a party in this case.
Document 2: the Michigan Model Criminal Jury Instructions, available at this PDF link: https://www.courts.michigan.gov/48d3b7/siteassets/rules-instructions-administrative-orders/jury-instructions/criminal/current/criminal-jury-instructions.pdf
Task:
Compare Document 1 against Document 2 instruction by instruction and line by line.
For every place where the filed jury instructions differ from the Michigan Model Criminal Jury Instructions, identify and list the deviation in table format.
For each deviation, provide:
1. Filed instruction number/title
2. Matching model instruction number/title
3. Exact filed text
4. Exact model text
5. Type of deviation:
– omitted language
– added language
– modified language
– reordered language
– merged/split instruction
– no matching model instruction identified
– model instruction omitted from filed set
– citation/reference error
– formatting only
– potentially substantive legal deviation
6. A brief explanation
7. A severity label:
– formatting only
– minor wording change
– potentially substantive change
– clearly substantive change
Rules:
– Use the PDF at the hyperlink as the source of the Michigan model instructions.
– Compare full text, not just headings.
– Flag every textual deviation unless it is purely spacing, pagination, or capitalization with no effect on meaning.
– Omit from the list all instances when the Brief explanation is Filed heading uses legacy ‘CJI’ prefix instead of current ‘MCrim JI’ prefix.
– Distinguish clearly between formatting differences and substantive changes.
– If the filed instruction paraphrases a model instruction instead of quoting it, treat that as a deviation and show both texts.
– If the filed set combines multiple model instructions, identify each model instruction involved.
– If a model instruction appears to be missing from the filed set, list it separately.
– Quote the relevant text exactly where possible.
– Group results by instruction number.
If the PDF link cannot be accessed or parsed completely, say so explicitly and identify any portion of the comparison that may be incomplete.
Output requirements:
Do not include source links, URLs, or inline citations in the response.
Return the final work as a downloadable .docx file and provide the download link. The final work must use at least 12.5-point font.
Here is the redacted testing sample I used for this example.
ChatGPT will create a helpful downloadable Word file report that looks like this. In this example, ChatGPT took 22 minutes in extended-thinking mode.
Special attention should be paid to the “Exact filed text,” “Exact model text,” and “Brief explanation” columns for flagged differences.
Both of Michigan’s model civil and criminal jury instructions are regularly updated—though the criminal instructions seem to be updated more often.
Enlisting ChatGPT can ensure (1) that you’re working with the current version of the model instructions, and (2) that those proposed by a party have been verified for accuracy against the current rules.