
OP-EXPOSÉ: The Rule of Communication Meets the Record
A JEDVIDENCE scan asks what an attorney’s work product communicates—and whether the documentary record can test it
By Bradley J. Burt | Outpost 422 / The Jaded Patriot Brief
There is a difference between accusing a lawyer of professional misconduct and examining legal work product through the rules governing the profession.
JEDVIDENCE is interested in the second.
My latest experiment begins with Wisconsin Supreme Court Rule SCR 20:1.4, the communication rule. It requires lawyers to keep their clients reasonably informed, reasonably consult about the means used to accomplish client objectives, respond to reasonable requests for information, and explain matters sufficiently to permit informed client decisions. Wisconsin’s current Rules of Professional Conduct remain contained in SCR Chapter 20. Wisconsin Court System
That does not mean Rule 1.4 gives me, as an opposing party, a communication right against another party’s attorney. It doesn’t. Instead, I am using the Rule as part of an experimental research framework: What happens when we take the communication concepts governing legal representation and use documentary metadata to examine the finished work product?
Exhibit A: The Position Statement
On May 19, 2026, attorney Storm B. Larson of Boardman & Clark submitted an eight-page position statement to Wisconsin’s Equal Rights Division on behalf of Camp Createability, LLC. The document identifies itself expressly as Camp’s response to my discrimination complaint, denies discrimination and retaliation, and asks ERD for a finding of no probable cause. Position Statement of Camp Crea…
That’s our starting point—not my opinion about Larson and not an AI characterization of Larson.
The document itself.
And the document makes propositions capable of being tested.
For example, the position statement says that on June 5, 2025, I mentioned retaliation and harassment to Emily Williams. According to the respondent’s narrative, Williams redirected me to CEO Debbie Armstrong. Position Statement of Camp Crea…
A few paragraphs later comes the proposition at the center of the retaliation defense: Armstrong supposedly had “no idea” I had mentioned retaliation or harassment and terminated me without knowledge of those allegations. Position Statement of Camp Crea…
The legal analysis then makes knowledge decisive. The position statement cites Gunty and Sabol, states that employer knowledge is required, and argues that the retaliation claim fails because Armstrong lacked knowledge of the protected activity when she made the termination decision. Position Statement of Camp Crea…
That is precisely where JEDVIDENCE stops reading like an ordinary reader.
It starts scanning.
Don’t Ask Whether the Lawyer Was Wrong. Ask What the Proposition Requires.
The proposition can be stripped of rhetoric:
Williams knew → Williams redirected Burt to Armstrong → Armstrong allegedly did not know → Armstrong made termination decision → lack of knowledge defeats claimed causal nexus.
Now we have variables.
When did Williams acquire the information? What exactly was communicated? What did Williams do afterward? With whom did she communicate? What information reached Armstrong? When did it arrive? What documentary artifacts memorialize those communications? Are there emails, texts, notes, phone records, witness accounts or contemporaneous documents? Does later work product describe the event differently?
Those are questions.
They are not findings.
The same technique applies to accommodation. The position statement acknowledges that Armstrong knew I occasionally experienced migraines while simultaneously maintaining that she did not know I possessed a specific disability requiring accommodation. Position Statement of Camp Crea…
Again, JEDVIDENCE doesn’t turn that difference into misconduct.
It turns it into a variable.
What did the employer know? When did it know it? From what source? What did “DVR client” communicate? What disability information was actually disclosed? What accommodation was actually requested? What documentation existed at the time?
Then we find the receipts.
Enter Generative AI
This is where the experiment gets particularly interesting.
Wisconsin legal-ethics guidance isn’t telling lawyers that generative AI itself is unethical. State Bar ethics guidance instead emphasizes that lawyers’ underlying professional responsibilities remain in place when they use AI. Wisconsin commentary has identified competence, confidentiality, supervision, verification, candor, communication and reasonable fees among the existing professional obligations implicated by GenAI. WisBar
And Wisconsin legal-ethics commentary has specifically warned against taking AI-generated legal documents “as-is.” Lawyers remain responsible for oversight and review of the resulting work product. WisBar
That principle works both directions in my experiment.
I don’t get to dump Larson’s position statement into ChatGPT and announce:
“AI says the lawyer violated the ethics rules.”
That would reproduce exactly the methodological problem I’m studying.
Instead:
SOURCE → AI SCAN → VARIABLE → PRIMARY AUTHORITY → DOCUMENTARY COMPARISON → HUMAN VERIFICATION → CORRECTION → CONCLUSION OR UNRESOLVED
If OpenAI misreads the position statement, I correct OpenAI.
If I misremember an event, the contemporaneous record gets to correct me.
If opposing counsel’s proposition survives comparison against the documentary record, it survives.
If the documents contradict my narrative, that contradiction belongs in the dataset too.
And if the evidence cannot answer the question?
UNRESOLVED.
That’s not a weakness. That’s source control.
SCR 20:1.4 Becomes a Microscope, Not a Verdict
The communication rule gives this experiment something more useful than a buzzword.
It gives us variables.
Consultation. Information. Explanation. Status. Decision. Knowledge. Response. Timing. Informed participation.
The State Bar’s AI guidance makes the connection even more interesting. Its 2024 discussion of generative AI explains that lawyers’ existing ethical responsibilities continue when GAI enters the representation, including communication and informed-consent considerations where appropriate. WisBar
So instead of asking whether artificial intelligence should replace a lawyer, JEDVIDENCE asks something much narrower:
Can AI help a human inspect whether legal work product is internally consistent with the documentary information available to test it?
That’s a very different proposition.
AI doesn’t become the witness.
AI doesn’t become the lawyer.
AI doesn’t become the judge.
And AI certainly doesn’t become the source.
It becomes the scanning instrument.
Where Subjective Coding Comes to Die
The Camp Createability position statement contains forceful characterizations. It describes me as responding in a “disrespectfully hostile manner,” says I “stormed” into a room, characterizes me as “visibly agitated and highly emotional,” and reports that others considered my behavior aggressive or frightening. Position Statement of Camp Crea…
Those words matter.
But a JEDVIDENCE scan doesn’t automatically accept them, and it doesn’t automatically reject them.
It asks what each characterization is made of.
Who observed the conduct?
When?
What precisely did the witness see or hear?
Was the characterization contemporaneous or retrospective?
Was there another witness?
Was an account written before or after the employment decision?
Did later versions change?
Does independent documentary evidence corroborate it?
That is what I mean when I say:
Keep the buzzword in the headline. Remove it from the finding.
The Gonzo journalist can describe how the accusation felt.
The researcher must dissect the variable.
The documentary record gets the last word it is capable of giving.
And the tribunal—not ChatGPT and not Bradley Burt—gets the legal decision.
The Experiment
That is the larger point of this project.
Generative AI does not eliminate human responsibility. It makes disciplined human responsibility more important.
Recent Wisconsin State Bar guidance continues to emphasize verification of AI-generated material, including citations. WisBar That is not an argument against the technology. It is an argument for building workflows in which mistakes become visible before they become finished work product.
My experiment therefore isn’t:
Can ChatGPT prove my case?
It is:
Can a human researcher surround generative AI with enough primary material, coursework, legal authority, documentary metadata and competing narratives that both human and machine errors become easier to detect?
That’s JEDVIDENCE.
And SCR 20:1.4 gives us another lens through which to conduct the experiment.
Don’t trust the machine.
Don’t automatically trust the storyteller.
Don’t automatically trust the advocacy document, either.
Scan the communication.
Extract the proposition.
Identify the variable.
Find the source.
Check the timestamp.
Preserve the competing account.
Correct the machine when necessary.
And then—
inspect the receipts.









