The Future Is Now: Why Trial Lawyers and Judges Should Embrace Generative AI Now and How to Do it Safely and Productively

[EDRM Editor’s Note: This article first appeared in American Journal of Trial Advocacy, Vol. 48: 323 (2025), and is published here with the permission of the authors: Ralph Artigliere¹ and Ralph C. Losey.² The opinions and positions are those of the authors.]

---

Introduction: The Urgency and Promise of Generative AI in Law

The unprecedented rapid advancement of generative artificial intelligence (AI) worldwide presents the legal profession with a pivotal opportunity for transformation. The legal system is deeply rooted in tradition, precedent, and established practices, which is good; however, this does not mean we should avoid technology. The legal practice must be open to change and embrace AI, just as it did with computers and online communications. We can keep our traditions of ethics, justice, and precedent, but also utilize generative AI to make our practices more efficient, consistent, and responsive to the demands of the modern world.

Trial lawyers and judges who embrace AI stand to gain substantial advantages in efficiency, productivity, and quality. The use of AI in legal applications across the profession is significantly increasing.³ Generative AI is already significantly improving legal research; drafting contracts, forms and pleadings; document reviews and document analysis; e-discovery; client communications; and the overall quality of the decision-making processes. Generative AI makes it possible for lawyers to obtain instant informal advice on expert issues, both factual and legal. Its application in legal practice is considered a game changer, as described in a White Paper prepared for the Arizona courts:

Imagine a system that can read, summarize, annotate, analyze, translate, categorize, synthesize, and interpret text and images at a post-graduate level. This is precisely what GAI is capable of today, and its abilities only continue to evolve. The implications for legal professionals are profound, and the use-cases are vast. AI has significant potential ranging from database searching, summarizing information, and generating correspondence. It can also assist with tasks such as contract drafting, policy compliance, data extraction, and due diligence. Litigation can also benefit from AI through e-discovery, legal research, deposition preparation, and drafting motions and pleadings. In essence, it’s transforming not just the practice of law but also the business of law. The automation of routine tasks has the potential to reduce the need for billable hours, a cornerstone of law firm profitability. Amidst the rapid change, AI emerges as a compelling solution to reinvent the legal industry for greater accessibility and efficiency.⁴

This does not mean machines will do the work for us.⁵ Lawyers and judges must still verify all AI output in the same way we have always verified the output of paralegals and young associates.⁶ The adherence to a high standard of ethics by our profession must remain. Generative AI is not about replacing the skilled minds of trial lawyers and judges; it is about enhancing their abilities.

The adherence to a high standard of ethics by our profession must remain. Generative AI is not about replacing the skilled minds of trial lawyers and judges; it is about enhancing their abilities.

Ralph Artigliere & Ralph Losey, The Future Is Now: Why Trial Lawyers and Judges Should Embrace Generative AI Now and How to Do it Safely and Productively, 48 Am. J. Trial Advoc. 323 (2025).

By leveraging AI as a partner, legal professionals can elevate their work and move beyond mundane, repetitive tasks, focusing their efforts on strategic, high-value activities and their unique human abilities.⁷ This hybrid⁸ co-intelligence approach—where human expertise and AI capabilities combine—can redefine the standards of legal practice, positioning lawyers and judges as more powerful and better informed decision-makers who are ready to navigate increasingly complex legal landscapes. Now is the time for trial lawyers and judges to take an active role in this transformation, embracing AI with confidence and clarity to shape a more efficient and future-ready practice.⁹

This Article guides trial lawyers and judges through the transformative potential of generative AI, demonstrating how to use it safely, effectively, and ethically.

I. Avoid the “Use Case” Mindset

To fully embrace the transformative potential of generative AI in the legal profession, it is essential to look beyond narrow, task-oriented applications and envision the broader, long-term impact. Generative AI is not just another tool to draft briefs faster or churn out judicial orders on demand. Instead, it represents a fundamental paradigm shift, offering legal professionals the opportunity to enhance their thinking, workflows, and overall effectiveness.

The allure of focusing on specific “use cases” is understandable. Questions like, “Can it write my brief in minutes?” or “Can it draft judicial orders effortlessly?” arise naturally. Yet, framing generative AI’s role solely in these terms risks reducing its transformative power to mere automation. The real value of generative AI lies not in its capability to do the work for you, but its potential to elevate how you work—empowering you to think more deeply, communicate more effectively, and achieve greater efficiency.¹⁰

Consider the evolution of technology in the legal field over the past few decades. Both of the authors of this article recall beginning their career in an era when every pleading was dictated to a legal secretary, typed manually, and corrected iteratively. The advent of automated typewriters, photocopiers, electronic research, fax machines, desktop computers, access to the internet, and, eventually, electronic filing, brought incremental improvements to the practice of law. Each step made legal work faster and more convenient.

But generative AI is not just an incremental improvement—it is a quantum leap. It offers legal professionals the ability to amplify their cognitive and analytical capabilities, creating a co-intelligent partnership that fundamentally redefines what is possible. Unlike previous innovations, which streamlined existing processes, generative AI has the potential to revolutionize how we approach legal analysis, case strategy, and communication.

Effective and timely communication is the core of lawyering, and generative AI strengthens that core in unprecedented ways. From enhancing clarity and precision in written arguments to generating insights that might otherwise be overlooked, the opportunities are vast. This is not about replacing lawyers; it is about augmenting their abilities and making them better in the areas they already excel.

Although we explore specific examples of how generative AI is being used in practice today, the real takeaway is this: generative AI is not just a tool—it is a partner. By working with it intelligently and ethically, you can evolve into a more proficient, adaptable, and forward-thinking legal professional. The question is not simply how generative AI can serve your immediate needs, but how it can reshape your approach to the profession itself.

II. Overcoming the Inertia of Traditional Law Practice

The promise of generative AI is enormous—an unprecedented efficiency boost that could handle mundane tasks, enhance legal analysis, and even augment the strategic thinking of legal professionals.¹¹ But with generative AI in its infancy, this promise is often buried beneath the clutter of naysaying, misunderstandings, and hype.¹² The legal headlines frequently highlight both the misuse of generative AI by lawyers and the resulting ethical opinions, court guidelines, and prohibitions.¹³ This narrative, combined with overblown claims from burgeoning tech start-ups, has contributed to a mixed atmosphere of opportunity and apprehension.¹⁴ This is an environment that adds to the drag from the natural inertia of our profession, built as it is on stare decisis and the slow evolution of case law and rules.

This resistance to change risks leaving legal professionals unprepared to meet the demands of a rapidly evolving technological landscape. Unlike the past, when slower societal and technological changes allowed practitioners to delay updates, today’s fast-paced environment demands continuous learning and adaptation. Embracing AI is no longer optional—it is essential for staying relevant and effective.

There is a way forward, a path built on study where lawyers and judges can keep pace with and embrace new technologies.¹⁵ Lawyers can capitalize on these new technologies and still maintain the traditions and high ethical standards of our profession.¹⁶ They can also avoid the missteps and setbacks of the reckless few in our profession who rush forward without the necessary learning and due care.¹⁷ By taking small but deliberate steps, lawyers and judges can transition to a future where generative AI becomes an ever more powerful ally; an intelligence tool that helps them focus on what matters most: effective advocacy, sound decision-making, and the delivery of justice.

This Article aims to break down the barriers that contribute to inertia, procrastination, and willful ignorance by providing guidance for integrating generative AI into legal practice. By understanding the technology and its applications, legal professionals can leverage AI in a way that complements their skills, respects ethical boundaries, and improves the quality of their work.

III. Understanding Generative AI

Artificial Intelligence has been around since the mid-twentieth century, but it has made remarkable progress in the 2000s with advancements in machine learning, the availability of big data, and powerful computing systems.¹⁸ This led to led to the advances in e-discovery with AI based active machine learning, also known as “predictive coding” document review methods.¹⁹ The predictive coding type AI methods were first approved by a court (Judge Andrew Peck) in 2012²⁰ in lieu of manual review. Remarkably, the approval has been uniformly followed by courts worldwide, although adoption of predictive coding methods by attorneys has been slow.²¹

On November 30, 2022 a new type of AI was released by OpenAI that used deep learning capabilities and neural networks whereby the AI began to teach itself successfully using Large Language Models (LLM) and large and powerful computer systems.²² Many companies quickly followed suit, including Google, who had invented the underlying technology but not applied it.²³ This change from active machine learning, which depended on human training and specified rules, to LLM self-learning systems, was a revolutionary breakthrough that caught most of the world by surprise. This new generative AI technology has since been adopted by hundreds of millions of users worldwide, influencing the arts, science, engineering, all types of business, education, academic research and professions such as law and medicine.

Less than a year after OpenAI’s launch, many lawyers and firms began exploring AI tools, as evidenced by the ABA’s 2023 Artificial Intelligence (AI) TechReport.²⁴ Most lawyers who were early adopters, Ralph Losey included, began using the early 3.5 version of ChatGPT as a writing tool.²⁵ A report from the Association of Corporate Counsel and Everlaw also highlights generative AI’s impact on corporate legal departments, with expectations of further transformation in in-house and outside counsel.²⁶ Despite this rapid adoption, the ABA survey revealed that many legal professionals are still concerned about developing competence in generative AI tools.²⁷

A better understanding of how generative AI and large language models (LLMs) work is essential to using them safely, accurately, and effectively. A fundamental explanation follows to provide greater clarity for practical use.

IV. How Generative AI Works

Large Language Models (LLMs) are advanced AI systems designed to understand and generate human-like text.²⁸ Think of them as exceptionally sophisticated text generators that can draft papers, answer questions, and hold conversations.²⁹ They are termed “large” because they are trained on vast amounts of text data, which helps them understand language patterns and context.³⁰ Training data is the fuel for LLMs.³¹ This massive collection of text from books, articles, websites, and other sources is processed to learn grammar, facts, and even some elements of common sense.³² The broader and more diverse the training data, the better the model can understand and generate coherent text.

Beyond the initial training, LLMs are subjected to fine-tuning with human input through advanced algorithms and programming.³³ Humans provide feedback on the model’s outputs, refining and adjusting responses to correct mistakes, improve accuracy, and align the model with desired outcomes.³⁴ This includes attempts to screen out biases contained in the underlying texts used for training.³⁵ Fine-tuning can also involve using specialized datasets, such as legal documents or medical records, to enhance proficiency in specific fields.³⁶

To simplify, Generative AI works by predicting the next word in a sequence based on the words that preceded it. Here is a simple breakdown:

Input: You provide the AI with a prompt or a question.
Processing: The AI analyzes the input using its training data to understand the context and predict the most likely continuation.
Output: The AI generates a response intended to be coherent and relevant to the input.

For example, if you ask Chat GPT about the requirements for privilege logs under Alabama law, the AI will use its training data to understand the question and generate an appropriate response. This might include references to procedural rules and cases that discuss privilege logs.

Here is what OpenAI’s GPT-4 with internet access might say if prompted: “What are the requirements for a privilege log under Alabama law?”:³⁷

The AI-generated response may appear well-crafted and accurate, and it contains links to its sources. Nonetheless, it is crucial to understand that such outputs always need to be reviewed and verified by a human.³⁸ Generative AI can sometimes make errors, and even worse it can sometimes produce confident sounding yet fabricated information, a phenomenon that has been called hallucination.³⁹ It is crucial to verify all output from generative AI just like you would verify what you receive from a law clerk or paralegal.⁴⁰

V. The Nature of AI and Limitations to be Aware of

Generative AI can be deceptively realistic in its conversational tone, but it does not reason or think like a human.⁴¹ It predicts each word by analyzing millions of data points to determine what comes next.⁴² Essentially, LLMs are extraordinarily advanced autocomplete systems, similar to the predictive text feature on your smartphone but trained to function at a near-human level of sophistication.

Essentially, LLMs are extraordinarily advanced autocomplete systems, similar to the predictive text feature on your smartphone but trained to function at a near-human level of sophistication.

Ralph Artigliere & Ralph Losey, The Future Is Now: Why Trial Lawyers and Judges Should Embrace Generative AI Now and How to Do it Safely and Productively, 48 Am. J. Trial Advoc. 323 (2025).

Although generative AI models like ChatGPT have demonstrated impressive capabilities—even passing the bar exam⁴³—it is vital to recognize their limitations. Understanding the technology allows users to harness its strengths and mitigate its weaknesses effectively.

A. No True Understanding or Reasoning

Generative AI lacks true comprehension. It does not “understand” text in the way humans do; it simply predicts what might come next based on patterns in the data it has been trained on.⁴⁴ This means that it can sometimes produce plausible sounding but incorrect or nonsensical answers.⁴⁵

B. Data Limitations and Bias

The output quality of generative AI is only as good as the data it has been trained on. Biases in the training data can lead to biased or flawed outputs.⁴⁶ Additionally, the model’s training data only includes information available up until a certain cut-off point, which means newer developments or recent legal changes might not be reflected.

C. Lack of Contextual Awareness

Generative AI does not have a genuine sense of context beyond the immediate input. It may struggle with nuanced legal concepts or long-term dependencies in complex arguments, potentially leading to errors that a human would easily catch.

D. Errors and Hallucinations

AI can generate highly detailed and specific responses that are factually incorrect—errors or even entirely fabricated hallucinations. Advances in platforms and careful prompting can mitigate the opportunities for both errors and hallucinations, but the potential for both cannot be totally eliminated.⁴⁷ Ralph Losey has performed extensive experiments on this subject and found that while errors can be reduced by careful prompting, they still continue, in large part because of errors and inconsistencies in the underlying data.⁴⁸ On the separate problem of entirely fabricated generative output—hallucinations—he has found them to be far easier to control and reduce by standard prompt engineering techniques.⁴⁹

E. Ethical and Confidentiality Concerns

Using generative AI for sensitive legal matters presents ethical challenges. Lawyers must ensure client confidentiality is not compromised, and the use of AI complies with all relevant ethical guidelines.

VI. Leveraging AI Effectively Despite Its Limitations: Human in the Loop

If legal professionals understand the fundamentals of how generative AI works, they can take steps to use it safely and effectively. To reach its practical utility, every output from these models must be subject to human review, especially in legal contexts. Proper use involves:

• Always verify. Confirm AI-generated responses against authoritative sources.
• Ethical Use. Respect confidentiality and avoid using AI for tasks that could compromise client privacy.
• Human Oversight. Employ a human-in-the-loop approach, where a professional supervises and refines AI outputs.

Generative AI presents an incredible opportunity for the legal profession to improve efficiency and expand capabilities. By combining human expertise with AI’s advanced processing, the potential for enhanced legal practice is vast—but only if used with care, knowledge,
and oversight.

Effectively integrating generative AI into legal practice requires adopting a human-in-the-loop approach. This means ensuring that all AI-generated content is carefully reviewed, verified, and refined by skilled professionals. By understanding AI’s strengths and limitations, legal practitioners can leverage it as a valuable tool to enhance their work, while maintaining accuracy, ethical integrity, and professional standards.

One popular generative AI platform, Claude by Anthropic, requires the human-in-the loop approach in its Usage Policy when using it for legal applications.⁵⁰ Anthropic considers certain uses of its product, including legal work, to be “high risk” because they “pose an elevated risk of harm because they influence domains that are vital to public welfare and social equity.”⁵¹ Accordingly, they require use of their products “related to legal interpretation, legal guidance, or decisions with legal implications” to be done with these “additional safety measures.”⁵²

Users should also always employ a thoughtful approach to the prompts they use with generative AI. Just chatting with AI in a haphazard manner can work sometimes, but for more complex applications it can be a recipe for frustration and failure. The machines require careful use of language to generate helpful results.⁵³ That is the purpose of prompt engineering. In the same way a good legal brief must be carefully crafted to produce the best results and comply with the requirements of the court, generative AI prompts must also be carefully crafted. This is where prompt engineering for lawyers comes into play, which has been the focus of Ralph Losey’s work with AI since 2023.⁵⁴

No competent lawyer or judge would take work product produced by a law clerk, paralegal, or associate and pass it off as their own without proper review.

Ralph Artigliere & Ralph Losey, The Future Is Now: Why Trial Lawyers and Judges Should Embrace Generative AI Now and How to Do it Safely and Productively, 48 Am. J. Trial Advoc. 323 (2025).

Importantly, viewing the need for human oversight as a limitation is misleading. No competent lawyer or judge would take work product produced by a law clerk, paralegal, or associate and pass it off as their own without proper review. The same principle applies to AI. When the tool is treated as a co-intelligence—offering immense capabilities for generating rapid work product—and the human user applies their own perspective, expertise, and judgment, both human and machine can perform at their greatest potential. The human-in-the-loop method is an essential component in the safe and effective use of AI.

The requirement to verify AI-generated results is a practical reality rooted in the inherent imperfections of AI in legal work, which demands precision, accuracy, logic, and the absence of bias. Accountability for output generated by or with the assistance of AI is non-negotiable under a lawyer’s legal and ethical responsibilities.⁵⁵ Ultimately, the “human” in the loop—the lawyer—bears accountability because it is the lawyer—not the AI—who is subject to regulation, professional standards, and the duty to serve clients effectively. Lawyers cannot cede this responsibility by blaming the machine for mistakes.

VII. Incorporating AI into Legal Work as a Co-Intelligence Partner—Rather than an “Easy Button” for Producing Work Product—Has Significant Benefits

AI can help identify and correct human biases and errors. Lawyers and judges, like all humans, are susceptible to bias and mistakes. By subjecting their work to generative AI, they can gain a broader perspective derived from vast training data, prompting them to reconsider and refine their views or arguments. The challenge in legal decision-making is not only ensuring access to all relevant information but also properly applying all facets of that information. Generative AI can serve as a valuable cross-check, enabling the “human” in the loop to confirm whether a decision is well-informed and unbiased. This is still a work in progress, however, because the AI itself has biases that are hard to detect and that may reinforce the human biases.⁵⁶

Ultimately, AI is merely a tool of the lawyer, albeit a more powerful tool than ever seen before. Lawyers must never abdicate their professional responsibilities and allow themselves to become a tool of the AI. Humans have capacities and talents that disembodied AIs may never acquire. Accountability will always rest on the human professional; however, by working in a hybrid manner with AI, man and machine can help mitigate each other’s biases and errors. This collaborative approach can ensure legal work maintains high standards of accuracy, ethical compliance, and professional responsibility.

VIII. Generative AI Programs and Platforms

The power and accuracy of generative AI have increased significantly since OpenAI introduced ChatGPT to the general public on November 30, 2022. Competition from other generative AI platforms like Claude by Anthropic, Gemini by Google (includes Deep Mind), Llama by Meta (Facebook), xAI (Elon Musk), Copilot by Microsoft,⁵⁷ and a few others have fostered rapid improvement and development. These “open-domain” or “general-purpose” AI platforms are designed to handle a wide range of tasks and queries across various domains, unlike “domain-specific” or “specialized” AI systems tailored for specific industries or applications, such as Lexis+ AI or Westlaw’s CoCounsel.⁵⁸

The major generative AI companies continuously update and train their models to improve their intelligence and abilities and to eliminate, or at least mitigate, the risks and errors. For example, when concerns about data confidentiality arose with ChatGPT 3.5—initially trained by learning from user inputs—OpenAI addressed this issue by modifying the terms of use for ChatGPT to allow users to purchase a license (starting at $20 per month) where the user “opts-out” of training and keeps user input confidential.⁵⁹ This privacy option and its equivalent in other models is significant for legal professionals and other users concerned about privacy. Lawyers have an ethical duty of maintaining the confidentiality of client data.⁶⁰ This issue highlights the need for lawyers and judges to understand the platform they are using and stay current with terms of use, just as they must do for other technologies like cloud computing, email, and messaging applications.⁶¹

Most AI products today have internet access, which allows them to provide up-to-date responses.⁶² Legal professionals need to understand the capabilities and limitations of each product because this is crucial to achieving optimal performance. These capabilities change rapidly, so constant diligence is recommended. Many law firms today either use outside specialists to advise them on AI related issues, including new products, or create their own AI departments with attorneys trained in this new field.⁶³

The available open-domain AI platforms vary in capability and characteristics. As competitors attempt to differentiate themselves, they offer variations in customization, accuracy, suitability for specific applications, and ease of use. Cost differences also exist, but at this point, the prices are quite reasonable, allowing users to utilize two or three platforms to best suit their needs and cross-check responses for improved output.

One remarkable feature of some generative AI products is their ability to be programmed or tailored for specific roles or functions through prompts or basic programming. For instance, OpenAI ChatGPT can assume personas assigned in prompts, such as a legal assistant, an eDiscovery lawyer, a judge, an image generator, or an editor. Additionally, custom GPTs can be created or accessed on OpenAI’s platform to perform specialized tasks. Ralph Losey, for example, developed a Visual Muse for OpenAI to create illustrations in various styles, as well as a GPT “Panel of Experts” to simulate consultations with a group of experts, providing diverse perspectives on an issue.⁶⁴

The potential applications of generative AI for legal and administrative tasks are both impressive and expanding as the tools evolve to meet the precision and accuracy required in law. Several use cases for these open domain tools are discussed in detail in subsequent sections.

IX. Specialized AI Systems

Westlaw’s AI system, CoCounsel, and LexisNexis’ Lexis+ AI are generally considered closed systems. This means they have secondary training on proprietary legal content and databases specific to their platforms, rather than open web content.⁶⁵ These systems are designed to provide accurate and reliable legal information by leveraging their extensive, curated legal databases instead of relying solely on the LLMs trained on internet and publicly available data, such as ChatGPT. Westlaw and Lexis as of November 2024 integrate OpenAI GPT-4 and other LLMs through their AI-assisted research tools.

Many generative AI applications today are based on a variation of this approach. They finetune the base generative AI model for a specific use case. This is called an AI application wrap.⁶⁶ This finetuning integration is designed to enhance legal research by providing quick, relevant answers to complex legal questions, as well as document drafting and summarization capabilities.

Although both Lexis and Westlaw platforms leverage the power of LLMs like GPT-4 to improve the efficiency and accuracy of legal research and related tasks, this presents both advantages and disadvantages. These systems constrain their AI by algorithms and training to be more precise, focusing their research on proprietary legal documents, including case law, statutes, legal journals, and copyrighted legal writing resources. The advantage of this approach is a reduction in the risk of generating fictional cases, a common problem in more open-domain platforms. However, errors in analysis and interpretation can still occur, underscoring the importance of verifying all outputs for accuracy.

One potential risk of using specialized systems is the misconception that a closed, specialized AI can generate a submission-ready brief without the need for verification or cite-checking.

Ralph Artigliere & Ralph Losey, The Future Is Now: Why Trial Lawyers and Judges Should Embrace Generative AI Now and How to Do it Safely and Productively, 48 Am. J. Trial Advoc. 323 (2025).

One potential risk of using specialized systems is the misconception that a closed, specialized AI can generate a submission-ready brief without the need for verification or cite-checking. Legal practitioners must remember that, although these systems are more precise because of their second level of proprietary training, they are still fallible.⁶⁷ Moreover, these specialized systems are often constrained by the very algorithms that make them reliable, which can limit flexibility and creativity.⁶⁸ In certain cases, thinking outside the box and pursuing creative advocacy are crucial for effective legal communication.

Specialized AI systems also include platforms designed specifically for certain law firm or court tasks, which incorporate generative AI to accomplish specialized legal functions. These programs are promoted as being designed to use organization-specific data and tailored to a firm’s specifications for workflow and security of input and output. Whether the products meet these lofty goals is up for continued debate just like the criticisms that have been lodged against Lexis and Westlaw products.⁶⁹ Examples of potential applications for generative AI platforms include eDiscovery search and analysis, privilege log creation, expert witness preparation and analysis, trial preparation, and creating exhibits for mediation or trial. If achieved, these applications of generative AI will allow law firms and courts to streamline complex legal tasks, enhancing efficiency and precision while remaining within a controlled, specialized environment.⁷⁰

Technology start-ups and existing players are filling the space with products tailored to a broad array of legal applications. These products cover everything from administrative and law firm management tools to products for eDiscovery, expert witness management, trial practice, contract creation and management, wills and estates, and much more.

X. Practical and Safe Applications for Lawyers and Judges

A. Research

Generative AI transforms legal research by enabling more nuanced, conversational queries, freeing practitioners from reliance on Boolean logic and older, pre-generative AI methods alone. Tools like Westlaw CoCounsel, Lexis+ AI, and Vincent AI streamline factual and legal research with greater efficiency.⁷¹ Success depends on selecting the right tool and crafting precise, context-rich prompts to guide AI toward accurate, relevant responses.

Generative AI platforms often provide direct answers supported by citations, streamlining the research process. Tools like Lexis and Westlaw, grounded in proprietary databases, offer enhanced reliability and integration with traditional research methods.⁷² These features enable practitioners to refine initial results and verify content with ease.

Factual and legal research can also be supplemented by the use of generative AI powered search engines, a/k/a Generative Search Engines (GSEs). GSEs revolutionize online research by delivering direct, conversational answers rather than extensive lists of website links. Unlike traditional search engines like Google, which require users to sift through hundreds of results, GSEs provide real-time, reliable insights tailored to specific queries. Though still evolving, GSEs promise to transform how legal professionals access and utilize information.⁷³

The authors foresee a day when traditional Google search is just used as a quality control step to verify nothing was missed by GSE and other Generative AI applications that may be used by attorneys such as Westlaw, Lexis, and vLex.⁷⁴ In other words, traditional link producing search may well become a secondary verification tool.

Despite its transformative potential, generative AI requires rigorous human oversight to ensure accuracy and reliability. Even the most advanced systems may miss nuances or introduce errors.⁷⁵ Over time, as AI technology evolves, these verification steps may become less burdensome, but they will remain essential. Legal professionals should view AI as a powerful supplement to their expertise, enhancing efficiency while upholding the diligence that rigorous legal research demands.

B. Emails and Other Routine Communication

Drafting client or opposing party emails demands precision and care, especially in the fast-paced environment of modern legal practice. Quick responses risk misunderstandings or unintended offense. Generative AI tools can assist by refining tone, structure, and clarity, ensuring professional and effective communication.⁷⁶ Generative AI excels in crafting precise, nuanced communication. By analyzing prior writing samples and contextual details, these tools can refine tone, adjust messaging, and enhance clarity, helping lawyers meet the demands of rapid communication without sacrificing professionalism.

Some email drafting software programs do not have professionalism guardrails and can write inappropriately harsh and offensive communications.⁷⁷ If you encounter this kind of AI generated language you should report it to the software company and complain. We have enough incivility in the profession already and do not need AI to fan the flames. If the software company does not respond, or worse, even encourages overly aggressive unprofessional content generation (as we have seen in early demonstrations by one major vendor), then we support reporting this to the legal community.

The professional advice to “sleep on” a response before replying to a volatile message remains sound, but using generative AI that has effective guardrails to refine your response can add an extra layer of safety and thoughtfulness.

Ralph Artigliere & Ralph Losey, The Future Is Now: Why Trial Lawyers and Judges Should Embrace Generative AI Now and How to Do it Safely and Productively, 48 Am. J. Trial Advoc. 323 (2025).

The professional advice to “sleep on” a response before replying to a volatile message remains sound, but using generative AI that has effective guardrails to refine your response can add an extra layer of safety and thoughtfulness. This process encourages careful reflection, ensuring your communication is effective and professional while minimizing the risk of misinterpretation or escalation. Remember, no matter who authored the email, you or your AI, it is still “your email” and the consequences will fall on your shoulders and that of your client.

C. Document Drafting

As communicators and advocates, trial lawyers often rely on written briefs as their primary tool of persuasion, especially in motions decided without oral arguments. In regard to these motions, clarity, conciseness, and relevance are paramount. Advocacy 101 holds that every word of a brief or memorandum should be relevant and consistent with the goal. Although some courts impose word or page limits, a good submission is always concise while remaining complete and coherent. Extra words can distract from the message or, worse, undermine the persuasiveness of the argument. Generative AI serves as a versatile writing assistant, refining arguments, reducing redundancies, and enhancing clarity. By acting as an editor, fact-checker, and persuasive writing partner, these tools enable lawyers to craft compelling and polished documents more efficiently.

LLMs not only draw from a broad range of excellent writing samples but are also adept at reducing redundancy and ensuring consistency and order in writing. Moreover, they allow for flexibility in adjusting styles, formality, and tone, which means that requesting a professional style can enhance messaging, making it more suitable for legal advocacy. Some specialized legal software boasts the ability to draft briefs and memorandums at the push of a button. However, relying on these products to produce a final draft is inherently risky.⁷⁸ Generative AI based software can be an excellent tool for generating ideas, structure, and preliminary drafts. But once again, human reasoning and oversight are essential to ensure the content is legally sound, logically coherent, and strategically aligned.

The same principles apply to judges who use generative AI platforms to draft orders or rulings. These platforms can be incredibly helpful when used properly, but human input, oversight, and the application of judicial reasoning are indispensable. Generative AI should be seen as an assistive tool—one that augments the work of skilled legal professionals rather than replacing their expertise. By using generative AI thoughtfully and ensuring proper oversight, legal professionals can harness its strengths while minimizing risks, ultimately producing more effective and polished legal documents.

D. Analysis & Summarization

Generative AI can analyze case data, summarize transcripts, depositions, and cases, and aid in trial preparation. Trial lawyers and judges can both benefit from crisp summaries. Summarization is available from all types of gen AI, although providers perform the tasks somewhat differently. For example, the generalized use platforms like GPT-4 and Claude provide richer, more complete case summaries while Lexis+ AI constrains its summaries to information that is precise and rather brief. Some platforms allow a choice between more or less comprehensive summaries. Lawyers and judges can choose the platform that best serves the needs for summarization. Many generative AI based software allows the user to request the product in different forms, such as unannotated summaries or summaries that are organized by issue or contain links to specific documents.

Trial lawyers understand the importance of messaging and nuance in negotiations with the opposing side, as well as in arguments before a court, jury, or arbitration panel. Words matter, and the way they are used can greatly impact the outcome of negotiations, mediations, and courtroom advocacy. Generative AI is particularly effective in this context because of its broad language skills and vast training, making it an ideal tool for crafting precise and targeted communication, refining terminology, and organizing messaging to ensure clarity and impact. Generative AI can also assist in verifying that analysis is consistent with the case theme, supports key evidence, and aligns with logical flow, comprehension, and the intended impact.

By analyzing the language used in negotiations or courtroom arguments, generative AI can help attorneys refine their messaging to maximize impact and better anticipate the intentions behind the language used by opposing parties. Additionally, AI can help adjust the tone and effectiveness of the language, ensuring it aligns with the desired level of formality and persuasiveness required in different legal contexts.

Lawyers and parties can enhance their effectiveness in negotiation, advocacy, and mediation, both in preparation and during sessions, by leveraging generative AI. These tools can assist in drafting compelling negotiation statements, organizing key arguments, and refining the language to ensure consistency and persuasive power. Generative AI can also help understand the subtle connotations of particular phrases and simulate potential responses from the other side, allowing legal professionals to prepare more strategically. The result is a more informed, consistent, and persuasive advocacy and negotiation process, ultimately aimed at achieving the best possible outcome for the client.

Effectiveness in presenting or challenging expert witnesses can make or break cases involving medical, scientific, or engineering testimony. Expert witnesses typically rely on a wide range of materials, including peer-reviewed journals, medical records, research studies, authoritative textbooks, diagnostic images, laboratory reports, guidelines from professional organizations, expert reports, and demonstrative evidence—all of which can be extensive and complex. Marshalling relevant factual information and testimony of witnesses, supplying this information to experts, and researching peer-reviewed materials, research studies, guidelines, protocols, and standards is challenging, and generative AI can assist in those tasks.

Past methods of compiling, reviewing, summarizing, and indexing these materials were often analog, making them expensive and time-consuming. Generative AI has an edge in managing large amounts of information and can provide summaries, analysis, and organization, turning a morass of complex materials into more manageable components. Whether preparing an expert to testify, preparing cross-examination of opposing experts, or developing final arguments or mediation presentations, the ability to quickly identify strengths, weaknesses, consistencies, and inconsistencies among various materials can be incredibly valuable and efficient.

Generative AI’s capabilities are particularly well-suited for this type of document-heavy and detail-oriented work. LLMs can assist in summarizing deposition transcripts, extracting key information from peer-reviewed articles, and providing quick, accessible insights into medical and scientific research. Furthermore, generative AI can help cross-reference information from diverse sources, ensuring the expert’s testimony aligns with the most current research and standards. Generative AI can also assist in organizing expert testimony and preparing or verifying demonstrative exhibits such as tables, charts, and diagrams, making complex information easier to present and understand.

This approach not only saves time but can also improve the quality of expert preparation and cross-examination. Generative AI can be used to simulate potential cross-examination questions, identify gaps or inconsistencies in an expert’s opinion, and ensure the arguments being made are well-supported by the relevant evidence. However, the role of human oversight remains crucial—legal professionals guided by consultation with experts must ensure AI outputs are accurate and contextually appropriate, and legal professionals must apply their own expertise to develop effective case strategies.⁷⁹

E. Translation Using Generative AI

Generative AI tools excel at quick, informal translations, making them useful for understanding phrases or converting text between languages.⁸⁰ Although not as accurate as specialized translation tools for formal work, they often capture nuances like slang or dialect, especially with context. These tools are particularly valuable in time-sensitive situations where immediacy outweighs the need for precision. However, translations should always be reviewed for accuracy before relying on them in critical matters.

F. Optimizing Legal Workflows with Generative AI

Generative AI is rapidly finding applications across both substantive legal practice and administrative tasks, with lawyers and their firms leveraging it to create efficiencies and save time. These advancements allow lawyers to focus on higher-value activities, such as strategic decision-making, complex problem-solving, and delivering high-quality work product. Similarly, in court systems, generative AI is streamlining clerical and processing tasks, contributing to improved efficiency and resource allocation.

To fully capitalize on the benefits of generative AI, it is essential to approach it with both knowledge and intentionality. Understanding its strengths and limitations is critical. Generative AI is most effective when thoughtfully integrated into workflows to enhance productivity and outcomes, rather than forcing workflows to conform to the constraints of specialized platforms. By learning how to wield this powerful technology appropriately, lawyers and courts alike can position themselves to unlock its full potential.

XI. Ethical Guidance: Navigating the Risks Responsibly

A. Competence and Ethical Obligations

The use of AI in legal practice demands adherence to established ethical standards, which require lawyers to maintain a level of competence that includes understanding new technologies relevant to their practice. Since 2012, the American Bar Association’s Comment 8 to Model Rule of Professional Conduct 1.1⁸¹ reinforces that competence involves staying abreast of changes in technology, including both the benefits and risks associated with its use, and the majority of states have followed that lead.⁸² Building on this foundational requirement, states like Florida have gone further by explicitly addressing generative AI in their ethical rules.⁸³

Florida’s Supreme Court recently enhanced the requirement of technology competence, specifically adding language to the Florida Rules of Professional Conduct emphasizing the responsibility of every lawyer to understand the benefits and risks associated with the use of generative AI.⁸⁴ The Comments to three other ethical rules were also amended to include specific requirements involving generative AI.⁸⁵ This evolution of the ethical rules reflects the recognition by The Florida Bar of the significant advantages of generative AI as well as the urgency of obtaining and maintaining the knowledge needed to use it properly in practice.

The ABA has issued a Formal Opinion on the use of generative AI tools that identifies some ethical issues and offers general guidance for lawyers based on the ABA Model Rules of Professional Conduct.⁸⁶ Bar associations, such as those in California,⁸⁷ Florida,⁸⁸ New Jersey,⁸⁹ New York,⁹⁰ Kentucky,⁹¹ Connecticut,⁹² and Pennsylvania⁹³ are developing ethical guidance and recommendations specifically aimed at generative AI. Like the ethical requirements they are based on, the guidance provides that legal professionals must be competent in their use of generative AI in order to capitalize on the power of the tools and to avoid potential pitfalls. For instance, the Florida Bar’s Ethics Opinion 24-1 stresses that lawyers must protect client confidentiality, provide accurate and competent services, and provide oversight of the generative AI tools.⁹⁴

Lawyers must “review the work product of a generative AI in situations similar to those requiring review of the work of nonlawyer assistants such as paralegals”⁹⁵ and to take full responsibility for work generated by or with the use of generative AI.⁹⁶ Lawyers have an ethical duty to ensure their staff use AI in conformance with the “lawyer’s professional responsibility and obligations.”⁹⁷ As for whether generative AI is an assistant for ethical purposes, the ABA changed the name of its rule on supervision from “Responsibilities Regarding Nonlawyer Assistants” to “Responsibilities Regarding Nonlawyer Assistance” in 2012.⁹⁸ The scope of the rule expressly entails the use of non-human assistance, such as a generative AI system or platform.⁹⁹ The California Bar’s guidance highlights the broad application of AI in various legal functions, but also the necessity for all practitioners to maintain their duties of competence and diligence when utilizing these tools.¹⁰⁰ Specifically, California’s guidance provides that

[a] lawyer must critically review, validate, and correct both the input and the output of generative AI to ensure the content accurately reflects and supports the interests and priorities of the client in the matter at hand, including as part of advocacy for the client. The duty of competence requires more than the mere detection and elimination of false AI-generated results.¹⁰¹

Determining the limits of reliance on generative AI will depend on the circumstances of the problem and the ability to verify the result. A recent prompt by Ralph Losey to ChatGPT 4o and its response is instructive: “Can you provide legal opinions?”¹⁰²

The rapid evolution of generative AI presents a moving target for lawyers, judges, and regulators. As the tools improve, so do best practices for their safe and effective use. Ignoring AI advancements in regulation and ethical guidance risks undermining a lawyer’s ability to serve clients efficiently.

B. Risk Mitigation: Errors and Hallucinations

Despite its potential, generative AI poses risks, including privacy concerns, inaccuracies, outright fabrications (referred to as “hallucinations”), and embedded biases. Rigorous human oversight is essential to mitigate these challenges and ensure reliable, ethical outcomes. Bar associations like the ABA as well as the Florida and New York State Bars emphasize the importance of ensuring AI-generated content is verified and validated before being used in any legal context. For example, Florida Ethics Opinion 24-1 provides that

a lawyer must review the work product of a generative AI in situations similar to those requiring review of the work of nonlawyer assistants such as paralegals. Lawyers are ultimately responsible for the work product that they create regardless of whether that work product was originally drafted or researched by a nonlawyer or generative AI. Functionally, this means a lawyer must verify the accuracy and sufficiency of all research performed by generative AI.¹⁰³

Inclusion of false information could have serious consequences for clients and lawyers alike.¹⁰⁴ Florida’s Opinion 24-1 focuses on verifying research performed by generative AI. Still, the requirement for verification extends to all work which involves generative AI, including drafting, summarizing documents or data, creating charts and exhibits from data sources, and so on.

Judges, law clerks, and court personnel have similar obligations. The Delaware Supreme Court Interim Policy on the Use of GenAI by Judicial Officers and Court Personnel reflects these concerns, emphasizing that generative AI should only assist, not replace, professional expertise.¹⁰⁵ This underscores the importance of using AI tools to support legal work while ensuring accountability and adherence to professional standards. With the stakes high and the landscape presenting a moving target, concern by lawyers and judges over the use of generative AI is understandable. However, the solution is not beyond our reach. In fact, it is quite manageable and rational as discussed below.

C. The Solution: Human-in-the-Loop

To make the best use of generative AI while mitigating its risks, users should employ a human-in-the-loop approach, which is described in section VI above. Any AI-generated content should be reviewed, verified, and refined by a knowledgeable human professional.¹⁰⁶ More accurately, for the legal environment, the responsible attorney must stand behind the accuracy of their work.¹⁰⁷ By understanding both the capabilities and limitations of AI, legal professionals can use it as a tool to enhance their work without compromising accuracy or ethical standards.

The need to verify results from AI is a practical reality rooted in the inherent imperfections of AI when applied to legal work, which demands precision, accuracy, logic, and the absence of bias.¹⁰⁸ Accountability for AI-generated outputs is essential, given the lawyer’s legal and ethical responsibilities.¹⁰⁹ Ultimately, the human in the loop bears accountability because it is the lawyer—not the AI—who is subject to regulation, professional rules, and the duty to serve clients effectively. Lawyers cannot absolve themselves of responsibility by blaming the machine for mistakes.¹¹⁰

Ultimately, the human in the loop bears accountability because it is the lawyer—not the AI—who is subject to regulation, professional rules, and the duty to serve clients effectively.

Ralph Artigliere & Ralph Losey, The Future Is Now: Why Trial Lawyers and Judges Should Embrace Generative AI Now and How to Do it Safely and Productively, 48 Am. J. Trial Advoc. 323 (2025).

The positive aspect of incorporating AI into legal work is that, when used as a co-intelligent partner for generating work product, AI can help identify and correct human biases and errors. Lawyers and judges are as susceptible to bias and errors as anyone else. By subjecting their work to generative AI, they gain a broader perspective based on vast training data, which may prompt them to reconsider and refine their views or arguments. The challenge in legal decision-making is not just obtaining all relevant information but ensuring that all facets of that information are appropriately applied. Generative AI can serve as a valuable crosscheck, enabling the human in the loop to confirm whether a decision is properly informed and unbiased.

Ultimately, accountability will always rest on the human professional.¹¹¹ However, by working collaboratively, humans and AI can mitigate each other’s biases and errors and enhance precision in communication. This partnership ensures legal work will maintain high standards of accuracy, ethical compliance, and professional responsibility. Lawyers must apply their expertise to validate AI-generated content and make necessary adjustments to meet the rigorous standards of the legal profession.

D. Risk Mitigation: Learn Prompt Engineering

Prompt engineering is a term of art that refers to the practice of crafting and refining inputs (or prompts) to effectively communicate with generative AI systems. Effective prompts are required to achieve accuracy and efficiency and to avoid hallucinations and mistakes.¹¹² Although some wrap type of software reduces the need for carefully crafted prompts to obtain useful responses, the mastery of prompt engineering skills remains crucial for effective and reliable use of generative AI.¹¹³

Generative AI systems are designed to be user-friendly, mimicking plain language human-to-human conversation, but achieving optimal results requires more than casual interaction.¹¹⁴ Unlike computer coding or Boolean search techniques, which many lawyers have used in legal research, prompt engineering is generally easier to learn but still demands a thoughtful and strategic approach.¹¹⁵

The success of communication with generative AI depends on crafting prompts that are detailed, clear, and logically structured so the AI can ingest, analyze, and respond appropriately.¹¹⁶ Those who use basic prompts like “Tell me about privilege logs” or “Write an email about preservation” are making the options for the tool too broad for useful responses. Simply asking for an answer or result without providing sufficient context or clear expectations will often yield suboptimal outcomes.¹¹⁷ Moreover, different generative AI platforms may interpret and respond to the same prompts in varying ways, making it crucial to tailor prompting strategies to the specific tool being used.¹¹⁸ Professor Josh Kubicki¹¹⁹ offers the following basic advice for prompting generative AI:

• Be intentional. Use words that reflect exactly what you want. Replace vague terms with specific ones.
• Provide context. Add a sentence or two to explain the purpose of your request. Is this a formal email? A witty social post? A technical report?
• Iterate. If the first output isn’t quite right, refine your prompt. Add examples or clarify your desired tone.¹²⁰

To achieve accurate and productive results, lawyers must develop proficiency in prompt engineering through experience and learning. Fortunately, this skill is accessible to lawyers, as it requires the foundational skills of logic, context-awareness, and analytical thinking—skills lawyers have already honed through traditional legal research and the use of automated tools. Professor Ethan Mollick has the following advice on learning prompt engineering, which is particularly apropos to lawyers and other professionals:

As it is a coworker, you want to work with it, not just give it orders, and you also want to learn out what it is good or bad at. Start by using it in areas of your expertise, where you are able to quickly figure out the shape of the jagged frontier of its ability. Because you are expert, you will be able to quickly assess where the AI is wrong or right. You do need to be prepared for it to give you plausible but wrong answers, but don’t let the risk of these hallucinations scare you off initially. Though hallucinations may be inevitable, you will learn where they are a big deal, and where they are not, over time.¹²¹

Although trial and error can help lawyers improve their prompting skills, a more efficient approach is to seek guidance from experts. Training in prompt engineering is available from a variety of sources, including generative AI vendors, online tutorials, and professional peers. Many AI platforms also offer built-in guidance for effective prompting,¹²² such as suggested prompts or interactive tips integrated into the tool’s interface. By embracing these resources, lawyers can rapidly build the skills needed to unlock the full potential of generative AI.

E. Risk Mitigation: Addressing Confidentiality Issues

In the use of any technology, such as cloud computing or even email, lawyers must understand the technology sufficiently to establish security measures that protect client confidential and law firm proprietary information.¹²³ Using generative AI platforms is no different. “The unauthorized access to, or the inadvertent or unauthorized disclosure of, information relating to the representation of a client does not constitute a violation” of ethical duties “if the lawyer has made reasonable efforts to prevent the access or disclosure.”¹²⁴ Reasonable steps for use of generative AI platforms include a thorough understanding of the terms of use for the platform with regard to handling of data and potential use or retention of data.¹²⁵

Providers of platforms designed for legal applications are sensitive to confidentiality requirements and provide for security. Nonetheless, understanding the tools and how to use them properly is the ethical responsibility of the user, such that careful inquiry of the provider and the development of safe use procedures are essential.

XII. How to Start Using Generative AI as a Co-Intelligent Partner

Every legal professional must determine their own pace for adopting AI, but comfort only comes with practice.¹²⁶ To truly unlock the value of generative AI, it is essential to start experimenting now.¹²⁷ These tools have the potential to significantly enhance both the quality and efficiency of legal work across various areas.

Begin by experimenting with free or low-cost generative AI tools, focusing on non-critical tasks. The best way to get started is by using AI for simple, routine tasks—such as drafting basic correspondence, summarizing documents, drafting simple contracts, or summarizing case law.¹²⁸ These low-risk applications provide a gentle introduction, allowing legal professionals to build skills and confidence without putting themselves at undue risk.¹²⁹ This allows the user to apply generative AI power to existing workflows rather than trying to adapt workflow to the tool. Such early experiments are valuable for developing familiarity with AI, which in turn reduces the sense of uncertainty and makes future adoption of AI tools less daunting.

Starting with small, manageable tasks helps lay the groundwork for more complex uses of generative AI as your comfort and competence with the tool grows. This low-risk introduction can foster familiarity and help legal professionals appreciate the potential of the technology. Many users will be surprised by the value AI can bring, and some may wonder why they did not start using these tools sooner. By taking these initial steps now, legal professionals can prepare themselves to be well ahead of the curve as generative AI tools become standard fixtures in law offices and courthouses.

For example, you might begin by using generative AI to draft an initial version of a routine legal letter, summarize case law, or create a summary of a lengthy deposition. These tasks can be reviewed easily, allowing the user to quickly assess the accuracy and usefulness of the AI-generated content. As familiarity grows, AI can be incorporated into more substantive and advanced tasks, such as analyzing legal precedents, organizing information for trial preparation, summarizing expert witness reports, or even critiquing the structure of an argument. The goal is to start small to build familiarity and capability incrementally, moving from low-risk applications to more complex, higher-value tasks.

Generative AI continues to improve daily, but it is not, and likely never will be, a fully automated solution capable of producing polished legal work without human oversight. There is no easy path to excellence in legal work.¹³⁰ However, AI empowers legal professionals to work faster, smarter, and more effectively. It can handle tasks that are time-consuming but relatively low-value, such as initial drafts or broad legal research, thereby freeing lawyers and judges to focus on the truly critical, high-value aspects of their work.

The most effective way to use generative AI as a co-intelligence partner is by embracing its strengths and understanding its limitations. As with any tool, AI works best when used thoughtfully and with clear human oversight. This means ensuring all AI-generated outputs are reviewed for accuracy, context, and ethical compliance. Generative AI should be seen as a powerful assistant—capable of enhancing the practice of law, but always in partnership with skilled legal professionals who bring judgment, ethical considerations, and critical thinking to the table. Every legal professional’s journey with AI will be different, and it is important to adopt these tools at a pace that feels manageable.

Conclusion: Charting the Future Together

The journey into generative AI is a partnership, not a replacement. As a co-intelligence partner, AI empowers lawyers and judges to streamline routine tasks and elevate their focus to the strategic and human-centered aspects of legal practice. By adopting AI thoughtfully and ethically, legal professionals can refine their work, expand their capacities, and redefine standards of excellence.

Early adopters of generative AI stand to gain significant advantages, positioning themselves as leaders in an evolving legal landscape. However, the responsibility for its use rests firmly on human professionals. By upholding ethical standards and maintaining rigorous oversight, lawyers and judges can harness AI’s potential while safeguarding the values at the heart of the legal profession.

Generative AI is not about abandoning tradition, instead it is about advancing the traditions of the legal profession.

Ralph Artigliere & Ralph Losey, The Future Is Now: Why Trial Lawyers and Judges Should Embrace Generative AI Now and How to Do it Safely and Productively, 48 Am. J. Trial Advoc. 323 (2025).

Generative AI is not about abandoning tradition, instead it is about advancing the traditions of the legal profession. By blending the best of human expertise with the power of AI, the legal profession can shape a more responsive, efficient, and equitable system of justice. The future of law is not a distant horizon—it is unfolding now, and it calls on each of us to lead the way.

Download a PDF version of the article here.

---

Notes

1. B.S., Eng. (1969), United States Military Academy, West Point; J.D. (1977), University of Florida School of Law. Judge Artigliere (retired), served as Circuit Judge for the Tenth Judicial Circuit of Florida after a twenty-four-year career as a trial lawyer. Artigliere is the editor of a five-volume LexisNexis Practice Guide on Florida Civil Trial Practice and the co-author of the LexisNexis book on Florida eDiscovery and Evidence, both of which he updates annually. Artigliere is on the Advisory Council on State and Local Judiciary for LexisNexis +AI, and he devotes most of his time to educating lawyers and judges on the intersection of law and technology. ↩︎
2. B.A. (1973), Vanderbilt University; J.D. (1979), University of Florida School of Law. Ralph Losey is Senior Counsel at Losey PLLC and CEO of Losey AI, LLC. Although he sometimes still serves as an Arbitrator in interesting cases, Losey now specializes in research, coding, and writing on generative AI as applied to law and life in general, not just discovery. He has authored over 100 articles on generative AI since 2023 and before that hundreds more on e-Discovery, plus five books published by the ABA and West. Losey also served as an Adjunct Law Professor at the University of Florida, where he designed the law school’s first accredited online course. ↩︎
3. See Ryan O’Leary, Generative AI in Legal 2024, RELATIVITY 4 (Nov. 2024), https://resources.relativity.com/generative-ai-legal-2024-infobrief.html (finding that AI use in the legal profession has increased by 43%). ↩︎
4. DARTH K. VAUGHN & WHITNEY L. STEFKO, THE UNSTOPPABLE WAVE OF AI: THE IMPERATIVE FOR ADAPTATION IN THE LEGAL INDUSTRY 4 (2024), https://www.azcourts.gov/Portals/225/AI%20and%20Practice%20of%20Law%20Final%20White%20Paper.pdf. ↩︎
5. THE STATE BAR OF CAL. STANDING COMM. ON PRO. RESP. AND CONDUCT PRAC. GUIDANCE FOR THE USE OF GENERATIVE A.I. IN THE PRAC. OF L.at 2, https://www.calbar.ca.gov/Portals/0/documents/ethics/Generative-AI-Practical-Guidance.pdf (last visited Nov. 22, 2024) (“[O]verreliance on AI tools is inconsistent with the active practice of law and application of trained judgment by the lawyer.”) [hereinafter California Practical Guidance]. ↩︎
6. See Fl. Bar Ethics Op. 24-1 (Jan. 19, 2024) at 4, https://www.floridabar.org/etopinions/opinion-24-1 (comparing the use of generative AI to the standards that apply to nonlawyer assistant’s and their competence). ↩︎
7. VAUGHN & STEFKO, supra note 2, at 5. ↩︎
8. See, e.g., Ralph Losey, From Centaurs to Cyborgs: Our Evolving Relationship with Generative AI, E-DISCOVERY TEAM (Apr. 24, 2024) (noting experts recommend two basic ways to use AI, both hybrid, where the unique powers of human intuition are added to those of AI), https://e-discoveryteam.com/2024/04/24/from-centaurs-to-cyborgs-our-evolving-relationship-with-generative-ai. ↩︎
9. Embracing generative AI offers more than just access to its many advantages. As generative AI continues to revolutionize business, government, and personal applications globally, lawyers who actively use and understand this technology will be better positioned to assist clients facing legal issues related to its use or misuse. True comprehension of technology is most effectively gained through hands-on application and practical experience. ↩︎
10. VAUGHN & STEFKO, supra note 2. ↩︎
11. Id. at 5-6. ↩︎
12. See id. at 9-10 (“Perceiving AI is a powerful colleague rather than a potential usurper is integral to [its] transformation [in the legal profession].”). ↩︎
13. Just a couple of months after generative AI burst on the scene in November 2022, a careless lawyer submitted a brief in federal court that was created by a colleague using Open AI’s Chat GPT 3.5. Neither lawyer verified the citations in the brief, some of which did not exist in reality. Mata v. Avianca, Inc., 678 F. Supp. 3d 443, 464-65 (S.D.N.Y. 2023). The lawyers in the Mata case were sanctioned—as they well should have been—but the case led to concern and some restrictions on the use of generative AI by some organizations and judges. See, e.g., Repository of Judicial Standing Orders Including AI Segments, EDRM—AI Ethics and Bias Group, https://edrm.net/judicial-orders-2 (last visited Nov. 23, 2024). For perspective on the pros and cons of restrictive court orders and guidelines, see Hon. Xavier Rodriguez, Artificial Intelligence (AI) and the Practice of Law, 24 SEDONA CONFERENCE J. 782, 783-84 (Sept. 2023); and Maura R. Grossman et al., Is Disclosure and Certification of the Use of Generative AI Really Necessary? 107 JUDICATURE 68, 72 (2023). Other lawyers have regrettably made the same mistake of not verifying the output of generative AI platforms. See, e.g., United States v. Cohen, 724 F. Supp. 3d 251, 254 (S.D.N.Y. 2024) (“Cohen had obtained the cases and summaries from Google Bard, which he ‘did not realize . . . was a generative text service that, like Chat-GPT, could show citations and descriptions that looked real but actually were not.’”). However, as discussed in detail below, generative AI can be used safely and effectively with appropriate human oversight. ↩︎
14. See VAUGHN & STEFKO, supra note 2, at 1 (“This . . . has naturally awakened intrigue within the legal community, particularly with the advent of applications like ChatGPT.”). ↩︎
15. See Grossman et al., supra note 11, at 69 (“GenAI applications are in widespread use, and billions of dollars are being invested in further development of this technology. The legal profession is not immune from these developments.”). ↩︎
16. See id. at 75-76 (providing a number of requirements that attorneys must comply with in order to ethically use generative AI in their practices). ↩︎
17. The lawyer in Mata failed to check the citations in a brief generated by ChatGPT, submitted it over his signature, and then doubled down by claiming the citations were real. 678 F. Supp. 3d at 450-51. Familiarity with generative AI and a modicum of care and professionalism would have avoided missteps and sanctions in this case. Every lawyer is ethically obligated to check the work of others, human or machine, who prepare a brief for them to be submitted over their signature. See Fl. Bar Ethics Op. 24-1, supra note 4, at 4. ↩︎
18. See Maura R. Grossman et al., The GPTJudge: Justice in a Generative AI World, 23 DUKE L. & TECH. REV. 1, 9 (2023). ↩︎
19. See Ralph Losey, TAR Course—1st Class: Background and History of Predictive Coding, E-DISCOVERY TEAM, https://e-discoveryteam.com/tar-course/tar-course-1st-class (last visited Nov. 25, 2024). ↩︎
20. See Moore v. Publicis Groupe & MSL Group, 287 F.R.D. 182, 183 (S.D.N.Y. 2012) (“This judicial opinion now recognizes that computer-assisted review is an acceptable way to search for relevant ESI in appropriate cases.”). ↩︎
21. Doug Austin, The Da Silva Moore Case Ten Years Later: eDiscovery Case Law,
    E-DISCOVERY TODAY (Feb. 23, 2022), https://ediscoverytoday.com/2022/02/23/the-da-silva-moore-case-ten-years-later-ediscovery-case-law. ↩︎
22. Luigi de Angelis et al., ChatGPT and the Rise of Large Language Models: the New AI-Driven Infodemic Threat in Public Health, FRONTIERS PUB. HEALTH (Apr. 25, 2023) https://pmc.ncbi.nlm.nih.gov/articles/PMC10166793. ↩︎
23. Ashish Vaswani et al., Attention Is All You Need, CORNELL U., https://arxiv.org/abs/1706.03762 (last modified Aug. 2, 2023) (Introduced the “Transformer” in 2017, a new neural network architecture that helped with language understanding. Before the Transformer, machines were not good at understanding the meaning of long sentences—they could not see the relationships between words that were far apart. The Transformer greatly improved this function and has become the basis of today’s most impressive language understanding and generative AI systems. OpenAI scientists immediately understood the significance of this discovery and began applying it right away. This led to its first release of a consumer Gen AI product, ChatGPT, whereas this insight and paper was not immediately considered for product use by Google.); see also Parmy Olson, Supremacy (St. Martin’s Press, Sept. 10, 2024). ↩︎
24. Darla Wynon Kite-Jackson, 2023 Artificial Intelligence (AI) Tech Report, AM. BAR ASS’N (Jan. 15, 2024), https://www.americanbar.org/groups/law_practice/resources/tech-report/2023/2023-artificial-intelligence-ai-techreport. ↩︎
25. See Ralph Losey, Chat GPT Helps Explains My Active Machine Learning Method of Evidence Retrieval, E-DISCOVERY TEAM (Jan. 28, 2023), https://e-discoveryteam.com/2023/01/28/chat-gpt-helps-explains-my-active-machine-learning-method-of-evidence-retrieval (providing example of knowledge of ChatGPT). ↩︎
26. GenAI and Future Corporate Legal Work, EVERLAW, https://www.everlaw.com/resources/acc-genai-report/#get-the-report. ↩︎
27. Kite-Jackson, supra note 22. ↩︎
28. Michelle Hawley, What Are Large Language Models (LLMs)? Definition, Types & Uses, CMSWIRE (Aug. 9, 2023), https://www.cmswire.com/digital-experience/what-are-large-language-models-llms-definition-types-uses. ↩︎
29. Id. ↩︎
30. Id. ↩︎
31. Id. ↩︎
32. Id. ↩︎
33. See Fine-Tuning Large Language Models (LLMs) in 2024, SUPERANNOTATE (July 23, 2024), https://www.superannotate.com/blog/llm-fine-tuning#:~:text=Large%20language%20model%20(LLM)%20fine%2Dtuning%20is%20the,models%20and%20turning%20them%20into%20specialized%20models. ↩︎
34. Id. ↩︎
35. Ralph Losey, Navigating the High Seas of AI: Ethical Dilemmas in the Age of Stochastic Parrots, E-DISCOVERY TEAM (Apr. 3, 2024), https://e-discoveryteam.com/2024/04/03/navigating-the-high-seas-of-ai-ethical-dilemmas-in-the-age-of-stochastic-parrots. ↩︎
36. Rachel Landwehr, Fine-Tuning an AI Model: What it Is, When to Use it, and How to Get Started, GOWAN COMMUNICATIONS (Nov. 29, 2024), https://gowancommunications.com/fine-tuning-an-ai-model-what-it-is-when-to-use-it-and-how-to-get-started. ↩︎
37. OpenAI, Response to: “What are the requirements for a privilege log under Alabama law?”, ChatGPT (Nov. 8, 2024), https://chat/openai.com. ↩︎
38. Ralph Losey, OMNI Version—ChaptGPT4o—Retest of the Panel of AI Experts—Part Three, E-DISCOVERY TEAM (May 9, 2024), https://e-discoveryteam.com/2024/05/29/omni-version-chatgpt4o-retest-of-the-panel-of-ai-experts-part-three. ↩︎
39. Id. ↩︎
40. Id. ↩︎
41. Id. ↩︎
42. Id. ↩︎
43. Ralph Losey, Bar Battle of the Bots—Part One, E-DISCOVERY TEAM (Feb. 26, 2025), https://e-discoveryteam.com/2025/02/26/bar-battle-of-the-bots-part-one (assessing the relative success of the latest advances in generative AI available in early 2025, advanced reasoning models, as compared to past efforts of AI to pass the Bar exam); Bar Battle of the Bots—Part Two, E-DISCOVERY TEAM (Mar. 5, 2025), https://e-discoveryteam.com/2025/03/05/bar-battle-of-the-bots-part-two (explaining how ChatGPT 4o reasoned through the legal issues before writing Bar exam answers). ↩︎
44. Losey, supra note 36. ↩︎
45. Id. ↩︎
46. Id. ↩︎
47. See id. (discussing one of the ways to reduce, but not eliminate, AI hallucinations). ↩︎
48. Id. ↩︎
49. Id. (“Importance of Context: AI models can generate plausible but incorrect outputs if the prompt or the context is not clearly defined. Legal professionals using AI tools need to provide clear, specific input to get the most accurate outputs, minimizing the risk of AI hallucinations.”). ↩︎
50. Usage Policy, ANTHROPIC, https://www.anthropic.com/legal/aup (last visited Jan. 6, 2025) (describing the human-in-the-loop approach as requiring content review by a qualified professional in the field). ↩︎
51. Id. ↩︎
52. Id. ↩︎
53. OpenAI Platform, Prompt Engineering, https://platform.openai.com/docs/guides/prompt-engineering (last visited Jan. 22, 2025) (explaining the importance of realizing that AI models need to be given precise instructions in order to give the best output). ↩︎
54. See EDRM’s AI Prompt Engineering Course for Legal Professionals by Ralph Losey, https://e-discoveryteam.com/pe-course-2 (last visited Jan. 6, 2024) (the course is now private but is expected to be open for enrollment by the time of this Article’s publication). ↩︎
55. ABA Comm. On Ethics & Pro. Resp., Formal Op. 512 (2024) (“Lawyers using GAI in litigation have ethical responsibilities to the courts as well as to clients.”). ↩︎
56. See Losey, supra note 33. ↩︎
57. Microsoft Copilot integrates OpenAI’s models into Microsoft products to enhance productivity within applications like Microsoft 365. Microsoft has made significant investments in OpenAI, totaling around $13 billion. This investment allows Microsoft to integrate OpenAI’s technology into its own products and services, and to be the exclusive cloud provider for OpenAI. See Microsoft and OpenAI Extend Partnership, MICROSOFT CORP. BLOGS (Jan. 23, 2023), https://blogs.microsoft.com/blog/2023/01/23/microsoftandopenaiextendpartnership. ↩︎
58. Lexis+ AI and Westlaw’s CoCounsel have large language models as their underlying generative AI engines. See Adam Allen Bent, Large Language Models: AI’s Legal Revolution, 44 PACE L. REV. 91, 126 (2023). ↩︎
59. See OpenAI Inc., Terms of Use, https://openai.com/policies/row-terms-of-use (last visited Jan. 21, 2025). ↩︎
60. ABA Comm. On Ethics & Pro. Resp., Formal Op. 512 (2024) (highlighting attorneys’ ethical duty of confidentiality). ↩︎
61. Id. ↩︎
62. See, e.g., Introducing ChatGPT Search, OPENAI (Oct. 31, 2024), https://openai.com/index/introducing-chatgpt-search. ↩︎
63. See Geoffrey D. Ivnik, Law Firms Hiring AI Talent Better Hurry, LEXISNEXIS: LEGAL INSIGHTS (Mar. 15, 2024), https://www.lexisnexis.com/community/insights/legal/b/thought-leadership/posts/law-firms-hiring-ai-talent-better-hurry; Gopal Ratnam, Law Firm Use of Data Scientists Grows Alongside AI’s Challenges, ROLL CALL (Nov. 22, 2024), https://rollcall.com/2024/11/22/law-firm-use-of-data-scientists-grows-alongside-ais-challenges. ↩︎
64. See Custom GPTs, LOSEY AI, https://losey.ai/custom-gpts (last visited Jan. 6, 2024); Ralph Losey, Panel of AI Experts for Lawyers, LOSEY AI, https://losey.ai/custom-gpts/evidence-that-ai-expert-panels-could-soon-replace-human-panelists-or-is-this-just-an-art-deco-hallucination (last visited Jan. 21. 2025). The custom GPTs are located on the Open AI GPT store and available for use without charge at https://chatgpt.com/gpts. ↩︎
65. CoCounsel and Lexis+ AI use LLM generative AI in their applications. The LLMs are trained on vast amounts of data outside of the content and databases of their platforms. The New Way to Work: CoCounsel, the GenAI Assistant for Professionals, THOMSON REUTERS (last visited Jan. 21, 2025), https://www.thomsonreuters.com/en/insights/articles/the-new-way-to-work-cocounsel-the-genai-assistant-for-professionals; Ray Collitt, AI Is Preparing the Next Generation of Lawyers, LEXISNEXIS, https://www.lexisnexis.com/html/ai-is-preparing-the-next-generation-of-lawyers (last visited Jan. 21, 2025); see Lexis+AI Frequently Asked Questions, LEXISNEXIS, https://www.lexisnexis.com/pdf/lexis-plus-ai-top-20-faq-sheet.pdf?srsltid=AfmBOoqZI-9Q27MywHC9ecqe7pZUJ9adyCmX3XbZIYtdCbD9OXaeewv (last visited Jan. 21, 2025); CoCounsel Is Powered by OpenAI’s GPT 4, the First AI to Pass the Bar, CASETEXT (Mar. 14, 2023), https://casetext.com/blog/cocounsel-powered-by-openai-gpt-4; GPT-4, OPENAI (Mar. 14, 2023), https://openai.com/index/gpt-4-research/?utm_source=google&utm_medium=paidsearch&utm_campaign=dsa&gclid=Cj0KCQiAqL28BhCrARIsACYJvkcyQLLgMC3s7BP1R0lx3EGYY_RV33yi-wpExyvAQjnk63o1ytRLcoEaAsrTEALw_wcB. ↩︎
66. See, e.g., How Do People Build AI Wrapper Apps?, REDDIT, https://www.reddit.com/r/SideProject/comments/18mla2c/how_do_people_build_ai_wrapper_apps (last visited Dec. 1, 2024). ↩︎
67. See Benjamin Perrin, Law Professor Gives Lexis+ AI a Failing Grade, CAN. BAR ASS’N (Nov. 12, 2024), https://www.nationalmagazine.ca/en-ca/articles/law/opinion/2024/law-professor-gives-lexis-ai-a-failing-grade; Caroline Hill, Canadian Law Professor Gives Lexis+ AI “A Failing Grade”—LexisNexis Responds, LEGALIT INSIDER (Nov. 18, 2024), https://legaltechnology.com/2024/11/18/canadian-law-professor-gives-lexis-ai-a-failing-grade-lexisnexis-responds. ↩︎
68. Adjustment of a platform’s creativity is a technical function of what is called the model’s “temperature” setting to allow more, or less, random selection of the predicted next word. The lower the setting, the more predictable and repeatable, but also the less innovative. Ralph Losey, Creativity and How Anyone Can Adjust ChatGPT’s Creativity Settings to Limit its Mistakes and Hallucinations, E-DISCOVERY TEAM (July 12, 2023), https://e-discoveryteam.com/2023/07/12/creativity-and-how-anyone-can adjust-chatgpts-creativity-settings-to-limit-its-mistakes-and-hallucinations; ChatGPT-4 Scores in the Top One Percent of Standard Creativity Tests, E-DISCOVERY TEAM (Sept. 7, 2024), https://e-discoveryteam.com/2023/07/21/chatgpt-4-scores-in-the-top-one-percent-of-standard-creativity-tests. ↩︎
69. See, e.g., Perrin, supra note 65; Varun Magesh et al., Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools, STANFORD.EDU (updated June 6, 2024), https://dho.stanford.edu/wp-content/uploads/Legal_RAG_Hallucinations.pdf. ↩︎
70. The Seminole County Florida court, together with the Clerk of Courts, created an AI driven system to identify incoming cases that meet and do not new Florida Supreme Court criteria. The system is said to be more accurate than human review of the incoming pleadings and it saves hundreds of hours of clerical work. The system was designed by an IT person who works for the Court to identify incoming cases that meet and those that do not meet certain criteria to comply with a new Florida Supreme Court reporting requirement. The system is said to be more accurate than human review of the incoming pleadings and it saves hundreds of hours of clerical work. The benefits asserted in this example are anecdotal, but many courts are increasingly adopting AI and other advanced technologies to improve efficiency and accuracy in case management and other judicial processes. These systems can help identify
    relevant cases, streamline document review, and reduce the workload on court staff. See, e.g., Jim Ash, AI Missteps in Court Filings Prompt Fla. Bar Rules Review, THE FLORIDA BAR (July 9, 2024), https://www.floridabar.org/the-florida-bar-news/ai-missteps-in-court-filings-prompt-florida-bar-rules-review. ↩︎
71. See, e.g., UC Davis Mabie Law Library, Generative AI Tools and Resources for Law Students, UC DAVIS SCHOOL OF LAW, https://libguides.law.ucdavis.edu/c.php?g=1386929&p=10257662 (last visited Jan. 22, 2025). ↩︎
72. See id. ↩︎
73. Ralph Losey, Generative Search Engines: Providing Answers Not Links, E-DISCOVERY TEAM (Sept. 6, 2024), https://e-discoveryteam.com/2024/09/06/generative-search-engines-providing-answers-not-links. ↩︎
74. Id. ↩︎
75. See Varun Magesh et al., AI on Trial: Legal Models Hallucinate in 1 out of 6 (or More) Benchmarking Queries, STANFORD UNIVERSITY HUMAN-CENTERED ARTIFICIAL INTELLIGENCE (May 23, 2024), https://hai.stanford.edu/news/ai-trial-legal-models-hallucinate-1-out-6-or-more-benchmarking-queries. ↩︎
76. See Phillip Tate, Generative AI: A Communications Professional’s Secret Weapon, SPINSUCKS, https://edrm.net/2024/09/generative-search-engines-providing-answers-not-links (last visited Jan. 22, 2025). ↩︎
77. See Shaun Chojnacki, Pros and Cons of Using AI for Email Writing, IPOST, https://www.ipost.com/pros-and-cons-of-using-ai-for-email-writing (last visited Jan. 22, 2025) (“[A]n AI system may misunderstand the context of an email and provide and inappropriate response.”). ↩︎
78. Jeffrey M. Allen & Ashley Hallene, TAPAs: Using Generative AI to Draft a Brief or Memorandum, AMERICAN BAR ASS’N (Apr. 29, 2024), https://www.americanbar.org/groups/gpsolo/resources/ereport/2024-april/tapas-using-generative-ai-draft-brief-memorandum. ↩︎
79. One of the valuable capabilities of generative AI is its ability to take on a persona or contextual perspective, simulating the expertise of different types of professionals. Thanks to the extensive training and diverse base of materials used in generalized platforms, it is possible to assign a persona to obtain advice, editing suggestions, and perspectives tailored to specialized expertise or specific contexts. For example, generative AI can adopt the persona of a federal judge, an eDiscovery lawyer, a law professor, an AI technologist, a general devil’s advocate, or even a ninth grader if a basic tone or simplified explanation is desired. See Losey, supra note 62. ↩︎
80. See generally 9 Best AI Translation Software (That You’ll Actually Use), COPY.AI, https://www.copy.ai/blog/ai-translation#:~:text=risk%2Dfree).-,2.,generate%20translations%20quickly%20and%20efficiently (last visited Jan. 17, 2025). ↩︎
81. MODEL RULES OF PRO. CONDUCT r. 1.1 cmt. 8 (AM. BAR ASS’N 2024), https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_1_1_competence/comment_on_rule_1_1. ↩︎
82. LexisNexis, Litigation Technology Competence State Law Survey, https://advance.lexis.com/api/document?collection=analytical-materials&id=urn:contentItem:67V1-PH81-JW09-M21G-00000-00&context=1000522 (last visited Nov. 21, 2024). ↩︎
83. See RULES REGULATING THE FLA. BAR r. 4.1-1 cmt. (Fla. Bar Ass’n 2024) (“To maintain the requisite knowledge and skill, a lawyer should keep abreast of changes in the law and its practice, engage in continuing study and education, including an understanding of the benefits and risks associated with use of technology, including generative artificial intelligence, and comply with all continuing legal education requirements to which the lawyer is subject.”). ↩︎
84. In re Amendments to Rules Regulating the Fla. Bar, 393 So. 3d 137, 139 (Fla. 2024) (per curiam) (“To maintain the requisite knowledge and skill, a lawyer should keep abreast of changes in the law and its practice, engage in continuing study and education, including an understanding of the benefits and risks associated with the use of technology, including generative artificial intelligence, and comply with all continuing legal education requirements to which the lawyer is subject.”) (emphasis added). ↩︎
85. Id. Requirements about lawyer responsibilities involving generative AI were also added to Rules 4-1.6 Confidentiality of Information (lawyers should act competently to safeguard information relating to the representation of a client by being aware that generative artificial intelligence may create risks to the lawyer’s duty of confidentiality); 4-5.1 Responsibilities of Partners, Managers, and Supervisory Lawyers (lawyers should establish internal policies and procedures to consider safeguards for the firm’s use of technologies such as generative artificial intelligence); 4-5.3 Responsibilities Regarding Nonlawyer Assistants (lawyers should consider requisite safeguards when assistants use technologies such as generative artificial intelligence). ↩︎
86. American Bar Association Standing Committee on Ethics and Professional Responsibility Formal Opinion 512- Generative Artificial Intelligence Tools, at 2, n.4, https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf. ↩︎
87. California Practical Guidance, supra note 3. ↩︎
88. Fl. Bar Ethics Op. 24-1, supra note 4. ↩︎
89. Jake Maher, NJ Courts Turn Attention to AI Impact with New Committee, LAW360 (Sept. 22, 2023, 4:56 PM EDT), https://www.law360.com/articles/1725012/nj-courts-turn-attention-to-ai-impact-with-new-committee. ↩︎
90. Report and Recommendations of the New York State Bar Association Task Force on Artificial Intelligence 29-32 (Apr. 2024), https://nysba.org/app/uploads/2024/04/Task-Force-on-AI-Report-draft-2024-04-02-FINAL.pdf. ↩︎
91. Kentucky Bar Association Ethics Opinion KBA E-457 (Mar. 15, 2024), https://cdn.ymaws.com/www.kybar.org/resource/resmgr/ethics_opinions_(part_2)_/kbae457artificialintelligenc.pdf. ↩︎
92. Christine DeRosa, Connecticut Bar Creates Panels to Examine AI, Women in Profession, LAW360 (Sept. 8, 2023, 4:35 PM EDT), https://www.law360.com/articles/1719738/conn-bar-creates-panels-to-examine-ai-women-in-profession. ↩︎
93. Pennsylvania Bar Ass’n Comm. on Legal Ethics and Pro. Resp. and Philadelphia Bar Ass’n Pro. Guidance Comm. Ethical Issues Regarding the Use of A.I., Joint Formal Op. 200 (2024), https://www.lawnext.com/wp-content/uploads/2024/06/Joint-Formal-Opinion-2024-200.pdf#:~:text=This%20opinion%20will%20clarify%20how%20our%20existing%20ethical,same%20extent%20they%20are%20in%20employing%20traditional%20methods. ↩︎
94. Fl. Bar Ethics Op. 24-1, supra note 4, at 1. ↩︎
95. Id. at 4. ↩︎
96. Id. ↩︎
97. MODEL RULES OF PRO. CONDUCT r. 5.3(b) (AM. BAR ASS’N 1983); Gary Marchant & Joseph R. Tiano, Jr., Artificial Intelligence and Legal Ethics, ARIZONA SUMMIT ON ARTIFICIAL INTELLIGENCE LAW AND THE COURTS (2024), https://www.azcourts.gov/Portals/225/AI%20and%20Legal%20Ethics%20Final%20White%20Paper.pdf. ↩︎
98. Marchant & Tiano, supra note 95. ↩︎
99. Id. Ultimately, lawyers are responsible for the quality and accuracy of work performed under their supervision, whether by human assistants or by technological tools like generative AI. Effective supervision requires thorough training, a clear understanding of the assistant’s capabilities and limitations, and diligent oversight to ensure the work meets the requisite standards of skill and competence. In the case of human and machine output, if the proposed work requires fact-checking or review of original source materials to verify accuracy, then it is the lawyer’s ethical duty to do so. ↩︎
100. California Practical Guidance, supra note 3, at 1-3. ↩︎
101. Id. at 3. ↩︎
102. OpenAI, Response to: “Can you provide legal opinions?”, ChatGPT, (don’t have date), https://chat/openai.com (enter query into “Message ChatGPT” box). ↩︎
103. Fl. Bar Ethics O. 24-1, supra note 4. ↩︎
104. See, e.g., Mata v. Avianca, 678 F. Supp. 3d 443 (S.D.N.Y. 2023) (describing two lawyers and their law firm being sanctioned following their use of false citations created by generative AI); Gutierrez v. Gutierrez, 2024 Fla. App. LEXIS 9374 (Fla. 3d DCA Dec. 4, 2024) (holding that a pro se party’s submission of fictitious case law to the court derived from ChatGPT was sufficient to warrant the imposition of sanctions). ↩︎
105. Delaware Supreme Court Interim Policy on the Use of GenAI by Judicial Officers and Court Personnel (Oct. 21, 2024), https://courts.delaware.gov/forms/download.aspx?id=266848 (“Generative AI tools are intended to provide assistance and are not a substitute for judicial, legal, or other professional expertise.”). ↩︎
106. This may involve becoming familiar with and consulting with technology experts as well as self-study. As always, lawyers need to understand their limits and get help or advice when needed. ↩︎
107. See Grossman et al., supra note 11, at 75 (noting the ethical obligation that attorneys have to provide accurate representations to the court). ↩︎
108. See id. at 73 (defining one of these inherent imperfections as “AI hallucinations”). ↩︎
109. As described in section X(A) above, ethical standards and many court rules and guidelines do not condone AI-generated legal outputs without human verification. See Fl. Bar Ethics Op. 24-1, supra note 4. ↩︎
110. See Mata v. Avianca, 678 F. Supp. 3d 443, 458, 464 (S.D.N.Y. 2023). ↩︎
111. See, e.g., Delaware Supreme Court Interim Policy on the Use of GenAI, Policy 1, supra note 104, at 1 (“Any use of GenAI output is ultimately the responsibility of the Authorized User. Authorized Users are responsible to ensure the accuracy of all work product and must use caution when relying on the output of GenAI.”); Fl. Bar Ethics Op. 24-1, supra note 4 (“Lawyers are ultimately responsible for the work product that they create regardless of whether that work product was originally drafted or researched by a nonlawyer or generative AI.”). ↩︎
112. See Ralph Artigliere, Words Matter: Tips on Effective Prompts to Improve Your Generative AI Output, EDRM BLOG (Feb. 13, 2024), https://edrm.net/2024/02/words-matter-tips-on-effective-prompts-to-improve-your-generative-ai-output. ↩︎
113. See Ralph Losey, Transform Your Legal Practice with AI: A Lawyer’s Guide to Embracing the Future, E-DISCOVERY TEAM (Jan. 24, 2024), https://e-discoveryteam.com/2024/01/24/transform-your-legal-practice-with-ai-a-lawyers-guide-to-embracing-the-future. ↩︎
114. Id. ↩︎
115. Id. ↩︎
116. Id. ↩︎
117. Id. ↩︎
118. Id. ↩︎
119. Josh Kubicki is a lawyer, entrepreneur, and teacher. Professor Kubicki is now a Visiting Lecturer at Indiana University Maurer School of Law where he teaches courses on legal operations and generative AI, among other subjects. ↩︎
120. Josh Kubicki, Brainyacts Newsletter No. 248 (Nov. 15, 2024), https://thebrainyacts.beehiiv.com/p/248-venture-capitalists-as-dc-lobbyists?utm_source=thebrainyacts.beehiiv.com&utm_medium=newsletter&utm_campaign=248-venture-capitalists-as-dc-lobbyists&_bhlid=d43718ddc5ebce8ab75e90085fb9c9161fbad603. ↩︎
121. Ethan Mollick, Getting Started With AI: Good Enough Prompting, ONE USEFUL THING BLOG (Nov. 24, 2024), https://www.oneusefulthing.org/p/getting-started-with-ai-good-enough [hereinafter Getting Started With AI]; see Ethan Goh et al., Large Language Model Influence on Diagnostic Reasoning A Randomized Clinical Trial, JAMA (Oct. 28, 2024), https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2825395?utm_campaign=articlePDF&utm_medium=articlePDFlink&utm_source=articlePDF&utm_content=jamanetworkopen.2024.40969 (discussing where GPT-4 was used in a randomized clinical trial including fifty physicians, and the use of an LLM did not significantly enhance diagnostic reasoning performance compared with the availability of only conventional resources). ↩︎
122. See, e.g., Prompt Engineering Overview, ANTHROPIC, https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/overview (last visited Nov. 24, 2024) ↩︎
123. Daniel W. Linna Jr. & Wendy J. Muchman, Ethical Obligations to Protect Client Data When Building Artificial Intelligence Tools: Wignmore Meets AI, AM. BAR ASS’N (Oct. 02, 2020), https://www.americanbar.org/groups/professional_responsibility/publications/professional_lawyer/27/1/ethical-obligations-protect-client-data-when-building-artificial-intelligence-tools-wigmore-meets-ai (last visited Jan. 26, 2025). ↩︎
124. MODEL CODE OF PRO. CONDUCT r. 1.6 cmt. (AM. BAR ASS’N 1983). ↩︎
125. Nicholas Daniel Seger, Understanding the Risk of Uploading Client Information to Generative AI Platforms, ABA (Jan. 16, 2024) https://www.americanbar.org/groups/young_lawyers/resources/tyl/practice-management/risks-uploading-client-information-generative-ai-platforms/?abajoin=true (last visited Jan. 26, 2025). ↩︎
126. Ethan Mollick, Thinking Like an AI, ONE USEFUL THING ON SUBSTACK (Oct. 20, 2024), https://www.oneusefulthing.org/p/thinking-like-an-ai?r=913wx&utm_medium=ios&triedRedirect=true (Professor Ethan Mollick suggests that, to introduce yourself to generative AI, use it for things you do for fun for about ten hours and you will learn “a remarkable amount.”); Getting Started With AI, supra note 119 (“The single most useful thing you can do to understand AI is to use AI.”). ↩︎
127. Mollick, supra note 124 (Be sure to follow any use policies of your organization.). ↩︎
128. See Artificial Intelligence Interim Guidance—AI and the Courts: Getting Started, NAT’L CTR. STATE CTS. (Mar. 2024), https://www.ncsc.org/__data/assets/pdf_file/0025/99232/RRT-AI-getting-started-march-2024.pdf (discussing how NCSC advises
     judges and court personnel to begin with a few low-risk internal facing tasks and documents before using AI tools on external facing items). ↩︎
129. Id. ↩︎
130. Owen Morris, The Transformative Power of Generative AI in the Legal Field, ABA (Sept. 12, 2023), https://www.americanbar.org/groups/law_practice/resources/law-technology-today/2023/the-transformative-power-of-generative-ai-in-the-legal-field. ↩︎

---

Assisted by GAI and LLM Technologies per EDRM GAI and LLM Policy.