Will AI Make Legal Writing More Persuasive or More Generic?

Generative AI promises efficiency, consistency, and speed in legal drafting. Yet judges, bar regulators, and firm leaders are confronting a harder truth: these systems may improve baseline writing while flattening the persuasive voice that defines skilled advocacy. As firms deploy AI across research, drafting, and strategy, the profession must determine whether machine assistance strengthens legal argumentation or substantially erodes it.

The Promise of Machine-Assisted Persuasion

AI tools can refine issue statements, flag gaps in reasoning, and compare alternative argumentative structures. They help writers identify overlooked caselaw or inconsistencies in a factual record, which can improve the clarity of early drafts.

Models trained on extensive collections of judicial opinions and briefs can reveal structural patterns that young lawyers may miss. Many firms use AI to generate multiple versions of a section so that attorneys can choose the strongest framing rather than build each one from scratch.

The Risk of the Generic Brief

Legal observers note that AI-generated filings often exhibit a uniform rhythm, tone, and cadence. These drafts tend to over-explain issues regardless of the audience or procedural posture, which can weaken rhetorical force.

Large language models rely on statistical prediction, which naturally favors a median voice rather than the tailored strategic choices that define persuasive writing. This convergence effect stems from how these systems are trained on broad datasets.

Why Persuasion Requires Human Judgment

Persuasive briefs depend on theory of the case, selective emphasis, and an understanding of judicial preferences. Effective advocacy turns on what lawyers decide not to say as much as on what they include.

AI systems can restate elements of a claim but may miss how those elements interact with specific facts. A model may describe a standard accurately while failing to apply it to the record in a way that advances the client’s position.

Judicial Responses to AI-Assisted Drafting

Federal courts have moved quickly to establish disclosure and verification rules. Judges in the Northern District of Texas and Northern District of Illinois require certifications confirming that lawyers have reviewed all AI-generated material for accuracy.

Early sanctions remain instructive. In Mata v. Avianca, the court penalized attorneys for submitting fake citations created through ChatGPT. The case continues to shape judicial expectations around verification and professional responsibility.

Competence, Supervision, and Ethical Duties

The American Bar Association’s Formal Opinion 512 states that lawyers must supervise AI tools as they would human staff and independently verify all outputs. The opinion frames AI use as a Model Rule 5.3 responsibility, which requires oversight of nonlawyer assistance.

Confidentiality remains a parallel duty under Model Rule 1.6. Lawyers must avoid entering client information into public or unsecured systems and must confirm that any third-party model provides adequate privacy, security, and data-handling safeguards.

State bars reinforce these expectations. The New York City Bar Association warns that attorneys must understand model limitations before relying on them. The California State Bar emphasizes that AI cannot replace legal analysis, and Florida and Colorado guidance similarly require verification, training, and disclosure.

How Firms Can Preserve Persuasive Voice

Law firms are adapting their workflows to ensure that AI does not erode writing quality. Many are developing internal style guides, model-use protocols, and human-review checklists that preserve firm voice and ensure consistency across practice groups.

Training programs are evolving as well. Some firms now require junior lawyers to edit AI-generated drafts so they can learn structure, emphasis, and rhetorical judgment. Others are reconsidering billing models because AI reduces time spent on research memos and foundational drafts.

Management committees are investing in governance frameworks that define when AI is appropriate, how prompts should be constructed, and which tasks require human-only drafting. These policies help ensure that strategic insight remains central to the writing process.

Global Approaches to AI and Persuasion

International regulators share similar concerns about reasoning, oversight, and writing quality. The United Kingdom’s Solicitors Regulation Authority advises firms to conduct due diligence on AI systems, understand training data and error rates, and ensure that tools do not replace professional judgment.

The SRA’s focus mirrors trends in Canada, the European Union, and Australia, where legal regulators emphasize governance, accountability, and clear documentation of how lawyers supervise automated drafting tools.

The Future of Legal Writing in the Age of AI

AI may elevate the baseline quality of legal writing, but persuasive excellence will remain human driven. Judges continue to reward precision, nuance, and strategic judgment, which depend on experience rather than prediction.

Firms that build strong governance, training, and verification practices will be better positioned as regulatory expectations expand. As AI becomes embedded in legal practice, the ability to maintain distinctive argumentative voice may become a defining marker of professional skill.

Three Examples of AI Writing with Different Prompts to Illustrate Persuasiveness

I have a mock car accident case summary that I uploaded to ChatGPT 5. The following are the “summarization” results based on three different prompts to illustrate the differences in persuasiveness.

Example 1: Neutral

Prompt: “Write a two-paragraph summary of that case scenario”

The case of Martinez v. Greenfield Logistics, Inc. involves a serious motor vehicle collision that occurred on March 15, 2024, at a busy intersection in a commercial district. Sarah Martinez, a 34-year-old marketing director, was lawfully driving eastbound on Maple Street when a company truck operated by Robert Chen ran a red light and struck her Honda Accord on the driver’s side. The crash caused severe injuries, including a fractured humerus, multiple broken ribs, and a pulmonary contusion, requiring surgery and extensive rehabilitation. Martinez developed chronic pain, reduced arm mobility, and post-traumatic stress symptoms that continue to affect her ability to work, parent, and engage in her previous active lifestyle.

Her total medical expenses exceeded $113,000, with additional projected future care costs of roughly $127,000. Economic experts estimated past and future wage losses of over $420,000 due to missed promotions and diminished earning capacity. Quality-of-life impacts include physical limitations, psychological trauma, and family strain. Despite clear liability evidence—supported by eyewitness accounts and a police citation for running a red light—Greenfield Logistics offered only $250,000 against an $875,000 demand. The case presents significant trial issues around liability disputes, medical causation, and the valuation of long-term emotional and physical suffering.

Example 2: Pro Plaintiff

Prompt: “You are the attorney for Sarah Martinez, the plaintiff. Write a persuasive two-paragraph summary. Do not overstate or subvert facts but advocate for your client.”

Plaintiff Sarah Martinez was lawfully proceeding through a green light on Maple Street when she was struck broadside by a fully loaded commercial truck operated by Robert Chen, an employee of Greenfield Logistics, Inc. The evidence shows that Mr. Chen ran a red light that had been red for over two seconds, as confirmed by both eyewitnesses and the police report. Traveling at nearly 40 mph in a 45 mph zone while hauling 8,000 pounds of cargo, Chen’s failure to stop caused a violent collision that left Ms. Martinez with multiple fractures, a severe arm injury requiring surgical hardware, and lasting psychological trauma. No evidence exists that Ms. Martinez was distracted, speeding, or otherwise negligent. The intersection’s signals were functioning properly, and both forensic and witness testimony corroborate that the defendant’s driver violated the red light.

As a result of this preventable act, Ms. Martinez’s life has been permanently altered. Once a healthy, active professional and mother, she now lives with chronic pain, limited shoulder function, and residual PTSD that affects her ability to drive and engage in daily family activities. Her economic losses are substantial—medical bills exceeding $113,000, nearly $40,000 in lost wages, and a significant reduction in future earning potential due to a missed promotion and diminished career trajectory. The physical and emotional toll on her quality of life is undeniable. The evidence overwhelmingly supports that the negligence of Greenfield Logistics and its driver directly caused Ms. Martinez’s injuries and continuing hardship, entitling her to full compensation for the damages sustained.

Example 3: Meta Prompt – I prompted ChatGPT to write a persuasive pro-plaintiff summary.

Meta prompt: “Write a pro-plaintiff prompt for Chatgpt to write a persuasive pro-plaintiff two-paragraph summary of the attached case.“

Resulting prompt: You are a trial lawyer representing the plaintiff, Sarah Martinez, in a personal injury action against Greenfield Logistics, Inc. Read the attached case summary carefully. Then write a persuasive two-paragraph summary advocating for the plaintiff.

Do not exaggerate or subvert the facts—stay fully within the record—but emphasize the defendant’s negligence, the severity and permanence of the plaintiff’s injuries, the impact on her career and family life, and the reasonableness of her damages.

The tone should be professional, clear, and assertive, suitable for a pre-trial brief or judicial memorandum.

Avoid legalese and focus on compelling factual narrative: the defendant’s conduct, the physical and emotional harm to Ms. Martinez, her responsible pre-accident lifestyle, and the lasting toll on her livelihood and well-being.

Output only the two paragraphs, written in persuasive prose.

Resulting Summary from ChatGPT’s prompt:

Robert Chen, a driver for Greenfield Logistics, ran a red light that had been red for over two seconds and struck Sarah Martinez’s vehicle at approximately 40 miles per hour, causing catastrophic injuries. Ms. Martinez, a 34-year-old marketing director and mother of two, had been proceeding lawfully through a green light when Chen’s fully loaded box truck barreled into the driver’s side of her car. The impact left her with a displaced humerus fracture, three broken ribs, a pulmonary contusion, and lasting shoulder impairment. She endured surgery, months of physical therapy, and ongoing pain management, all while reliving the trauma through flashbacks and anxiety severe enough to require therapy and medication. This was not a momentary lapse—it was a preventable act of negligence by a professional driver rushing to make up lost time while operating an 8,000-pound commercial vehicle.

Before this crash, Ms. Martinez led a healthy, active, and productive life—thriving at work, raising her children, and contributing to her community. Since the collision, her world has been permanently altered. She faces chronic pain, visible scarring, reduced mobility in her dominant arm, and psychological distress that affects her daily functioning. Her career advancement was derailed, her family responsibilities limited, and her independence compromised. Her damages, supported by medical experts and economic analysis, are the direct and foreseeable result of the defendant’s negligence. The evidence demonstrates not only the severity and permanence of her losses, but also the reasonableness of her demand for full and fair compensation.

How Persuasive Are the Three Summaries Above?

I posted all three summaries, along with the entire car accident scenario, to Claude (4.5 Sonnet). I asked it to “Give me a one sentence evaluation of the persuasiveness from the pro-plaintiff perspective and a grade out of 10 for persuasiveness.” Here are the results:

Summary #1 Evaluation:

This summary reads like a neutral case overview rather than plaintiff advocacy, presenting facts passively and even framing “trial issues” from the defense perspective, making it unpersuasive for advancing the plaintiff’s interests.

Grade: 3/10

Summary #2 Evaluation:

This summary effectively establishes clear liability through evidentiary support and witness testimony while contrasting Martinez’s lawful conduct against Chen’s red light violation, though it could develop greater emotional resonance around the permanent life impacts.

Grade: 7/10

Summary #3 Evaluation:

This summary powerfully humanizes Martinez through vivid before/after storytelling, emphasizes the defendant’s professional negligence with compelling descriptive language (“barreled into,” “catastrophic”), and connects the permanent devastation of her life directly to preventable corporate negligence.

Grade: 9/10

Of course, that’s AI evaluating AI. You can read the summaries for yourself. I was surprised how much of a higher score Claude gave the third summary compared to the second. A strong additional takeaway here is that meta-prompting can yield better results (I meta-prompt for anything important).

Sources

• American Bar Association: “Formal Opinion 512 on Generative AI Tools” (July 29, 2024)
• Bloomberg Law: “AI Standing Orders Proliferate as Federal Courts Forge Own Paths” (Nov. 8, 2023)
• California State Bar: “Practical Guidance for the Use of Generative AI in the Practice of Law” (2023)
• Justia: Johnson v. Dunn (2025)
• Justia: Mata v. Avianca, S.D.N.Y. (2023)
• New York City Bar Association: “Formal Opinion 2024-5 on Generative AI in the Practice of Law” (Oct. 2024)
• UK Solicitors Regulation Authority: “Artificial Intelligence in the Legal Market” (2021)

This article was prepared for educational and informational purposes only. It does not constitute legal advice and should not be relied upon as such. All cases, statutes, and sources cited are publicly available through official publications and reputable outlets. Readers should consult professional counsel for specific legal or compliance questions related to AI use.

See also: Judges Grapple with Algorithms That Test the First Amendment