21st February 2024

“In recent years, artificial intelligence (AI) has gained significant attention in various industries, and the legal profession is no exception. AI has the potential to revolutionize the way lawyers work, from automating mundane tasks to assisting with complex legal research and analysis. The use of AI in the legal field has sparked debate about its impact on the role of lawyers and the future of legal work. This article will explore the current and potential applications of AI in the legal profession and its potential benefits and challenges.”

– ChatGPT, in response to the prompt, “Write an introductory paragraph on an article about artificial intelligence in the legal profession.” The capabilities of AI are truly impressive.

ChatGPT has been making headlines. On March 21, 2023, The Wall Street Journal reported that Google opened to the public its conversational AI program, named “Bard.” The world is now seeing a “space race” in the world of AI. But how will this affect the legal field? Will these AI programs be developed to draft briefs, do research, etc., autonomously? Will it replace lawyers? Can AI be used to make legal services cheaper?

It is still too early to predict the future of all places where AI can be used. We can surely expect, however, that AI will be (and arguably has already been) used in the legal field, but only as tools to be used by attorneys, and not to replace attorneys.

An often-asked question is whether AI will replace lawyers. There are obvious ethical hurdles for supplanting lawyers with AI programs. For instance, the Model Rules of Professional Conduct require that a lawyer “provide competent representation to a client. Competent representation requires the legal knowledge, skill, thoroughness and preparation reasonably necessary for the representation.”[1] It is difficult to imagine that a computer program, no matter how sophisticated, could “represent” a person. Lawyers act as intermediaries between the parties, a role that requires more than legal knowledge: it requires human understanding, compassion and wisdom. That benefit dissolves if AI replaces lawyers. In addition, each state requires that lawyers be admitted in that state and maintain Continuing Legal Education requirements. Clients often think they know what they want, but sometimes lawyers can better advise them on what results are possible. Also, lawyers are accountable to their clients. If a client is unhappy with the lawyer’s representation, they can go to the state bar to seek redress. If lawyers are replaced by AI, the question of who is accountable becomes more difficult — the company who developed the software? So, to answer the question of whether AI can replace lawyers: No.

Perhaps the better question is whether AI can be a tool that non-lawyers can use to perform legal tasks on their own. This question is reminiscent of a story that has been attributed to George Washington. The story goes that, after his presidency, George Washington was asked to resolve a land dispute between two adjacent landowners about the location of a property line. Washington surveyed the property, determined the proper property line and sent an invoice for $100 to the landowners. The landowners sent the invoice back requesting that Washington provide more detail. So, Washington sent: “$10 for a metal stake, $90 for knowing where to put it.”

No matter what tools are available to individuals who want to handle the matters themselves – AI or any other technological advancements like legal research programs – there will always be a need for the analytical skills of lawyers, including not only knowing the answers to certain questions, but knowing what other questions to ask. It will always be necessary to perform a final check on the documents being submitted to the court or to opposing parties to ensure something has not gone wrong in the use of the AI program.

When Westlaw and LexisNexis first developed their online legal research databases, there were likely similar questions about whether this advancement would replace lawyers. Instead, what happened is it freed attorneys to spend more time on substantive work rather than being stuck in law libraries researching and “shepardizing” cases to support or refute legal issues. With these tools, lawyers could do their jobs more thoroughly and more quickly, thus saving time and money for the client. The same is true with AI.

If AI could take on some of the more mundane tasks of lawyers, it may free them up to better analyze their client’s situation, better develop strategy and allow them more time to refine their work product – providing a better result for their client.

It is undisputed that the speed of AI’s progression has sparked concerns and conversations as to how AI will be used in the future and whether such programs will be allowed to go unchecked by humans. In the legal field, such curiosities have increased due to AI’s recent passing of a bar exam.

The bar exam serves as a gatekeeper to the legal profession by testing people’s capability to practice law competently in that state. Most lawyers will agree that the exam stands as one of the most agonizing mental marathons for attorneys, beginning with day-long bar prep classes and extending to studying hours every day for months, all after soldiering through three years of law school. The bar exam requires vigorous preparation. Each state bar exam has multiple essay questions and, in most cases, 200 multiple-choice questions (commonly referred to as the MBE) drafted by the National Conference of Bar Examiners. The MBE often requires the most rigorous training, with scores ranging from 40 (low) to 200 (high).[2] The median score for the MBE is 140.7.[3]

Near the end of 2022, researchers tested ChatGPT by having it take a bar exam. ChatGPT failed the exam. Around four months later, a new and improved Microsoft-backed AI program, GPT-4, passed the bar exam with exceptional scores.[4] The program answered nearly 76% of multiple-choice questions correctly, which surpasses the average human test-taker by about seven percent.[5] The program scored a 297 out of 330 on the entire exam, including answers to the essays, ranking in the 90th percentile of human test-takers.[6]

GPT-4’s ability to rapidly progress from failing the bar to passing it in the 90th percentile in approximately four months is another reason for concern in the legal profession regarding replacement. But, in its current state, AI lacks the strategic skill to replace a licensed human attorney. Lawyers are trained to think analytically and strategically. Being a lawyer requires human engagement and human creativity. Law professors Deborah Jones and Mary Lu Bilek explain that even as AI’s bar-passage rate increases as it is exposed to more legal materials, human value cannot be replaced. At the very least, human beings will be required for “knowing when to consult AI, how to assess AI responses and how to integrate AI knowledge with the human dimensions of a client problem.”[7] The professors reiterate that it is the quality of the human lawyer that the profession requires throughout each unique practice. AI will always need a lawyer to give it instructions on what to create and what information to use to create it. At most, AI is expected to assist lawyers in the legal field, not replace them.

One benefit that derives from AI’s capability in the legal profession is assistance with legal research. In fact, AI is already present in legal research platforms such as Westlaw, LexisNexis and Google Scholar, where natural language searches are incorporated into their databases. Natural language searches eliminate the need for awkward Boolean search queries. Now, lawyers can simply ask a question in the search box and get results. For example, rather than searching ‘“elements” /s “battery”’ and sifting through potentially thousands of results, researchers can ask, “What are the elements of battery?” The implementation of plain language searches has decreased the time it typically takes to research relevant issues, allowing lawyers to put that research into work product more quickly and save legal expenses.

Even with research tools, a human attorney is required to understand the law and how it fits into the big picture for the client. Sometimes creativity is needed when relevant case law is lacking and analogies must be drawn. In addition, the researching attorney must decide which opinions should be used and which should not. Attorneys often consider the facts that led to the decision, the jurisdiction and the posture of the case, as well as the opinion’s author and its use as precedent or persuasive authority elsewhere. The research process is carried out with an understanding not only of the client’s case but also the client’s ultimate objective, which may be more than just to “win.” This creativity and need for flexibility cannot be matched by a software program, no matter how sophisticated.

 Overall, legal research tools have not eliminated the value a human attorney brings to the case. Rather, they show the ability to assist in completing legal tasks at a more efficient rate.

No matter how sophisticated the software becomes, it will still need a human attorney to direct it, edit it and approve of the final product.

AI best fits in the legal field as a tool to be used by attorneys. Beyond research capabilities, AI can be further employed to draft simple pleadings and documents, leaving attorneys to focus on strategy in representing their clients and relying on their experience, judgment and wisdom.

For instance, asking ChatGPT to draft a Non-Disclosure Agreement between Company X and Company Y results in a standard form for a Non-Disclosure Agreement. It is a good place to start so that an attorney does not have to start from scratch. However, an attorney, based on experience, may recognize that there are certain documents that need to be protected at a higher level than others. The attorney can advise their client of additional provisions to be included in addition to the form that was provided by the ChatGPT program. The ChatGPT results cannot replace the work of attorneys in addressing nuances specific to each client, editing it and making it better.

Perhaps, one of the most beneficial uses of AI is for the underserved population. Frequently, this population needs assistance with uncomplicated pleadings and documents, such as Wills and Advanced Directives and No-Fault Divorces. If attorneys can use these tools to be more prolific in drafting such pleadings and documents, then their time can be used to better determine from their client what is needed and use the tools to create them.

No matter what, AI is being developed and refined. The world can expect better and more useful versions. Attorneys would be best advised to welcome these advancements and find productive and ethical ways to incorporate them into their practice. Attorneys can start by accessing ChatGPT or Bard, asking some questions and seeing what it can do for them.


[1] Model Rule of Professional Conduct, Rule 1.1.

[2] https://www.ncbex.org/exams/mbe/scores/#:~:text=How%20Is%20the%20MBE%20Scored,)%20to%20200%20(high)

[3] Id.

[4] https://www.jdjournal.com/2023/03/16/revolutionary-bar-exam-results-ai-matches-human-lawyers-scores-say-researchers/

[5] Id.

[6] Id.

[7] https://news.bloomberglaw.com/business-and-practice/chatgpt-almost-passed-the-bar-but-competent-lawyers-do-much-more

Finis

link