AI is a Lawyer’s Technician, Not a Replacement

After receiving a copy of the initial complaint filed against her client, the attorney hands it off to a robotic assistant. The assistant first provides the attorney with a summary of the current law on the specific subject matter raised by the complaint, and provides a few key cases from her jurisdiction. As the attorney begins following up on those research leads, her robotic assistant gives her a report on the estimated value of the case based on venue, judge, opposing counsel, and past settlements and awards in cases with similar facts. She’ll use this report in her client meeting this afternoon as they discuss the client’s potential exposure, decide if they should make a settlement offer, and so on. As the attorney completes her preliminary research, her robotic assistant provides her with an initial round of litigation documents it’s drafted for the case. The attorney reviews these documents, which include a preliminary answer and a request for production and admissions, and finds that only a few minor corrections are necessary before she is comfortable signing and filing them. By the time she finishes her morning coffee, she has already completed a day’s worth of research, and has a ready-to-file answer and a set of initial written discovery in hand. This is what the morning of an attorney who has integrated AI into every aspect of her practice could look like.

None of this is speculative: as of this writing artificial intelligence applications, either in current use or in the final stages of development, are capable of doing each of the tasks that our fictional robotic assistant performed. Examples include Ross Intelligence and CaseText AI-based legal research aides; Premonition Analytics, an AI-based case valuation service; LegalMation, for automated document drafting. Understandably, these developments prompted some existential fear that any part of our profession, which requires extensive education and rigorous certification, could be automated. However, this fear is unfounded; as our fictionalized attorney and her robotic assistant illustrate, AI will not replace attorneys, but instead will enhance their individual capabilities in ways that will benefit attorneys and clients alike.

The legal services industry faces some unique challenges that artificial intelligence is uniquely suited to solving. To explain why, a definition of terms is first necessary. Knowledge-workers, such as attorneys, accountants, engineers, programmers, etc., can be generalized into two basic categories: experts and technicians. A technician is retained for specific skills, and is paid to perform specific tasks—often ones that person has done many times before. Conversely, an expert is retained to provide a broad base of knowledge and to use that knowledge to synthesize new solutions to problems that have never been seen before. A technician requires less education, but generates value for an employer or client every hour spent working. An expert is comparatively costly to train, but generates value for a client only when that expertise is used to synthesize a new solution, and not every application of that same solution thereafter.

Given these two definitions, it’s clear attorneys are experts by training. In law school, they are taught the doctrines that underpin the law, rather than the specifics of the law itself. Indeed, many attorneys will tell you that the real value of law school is being taught “how to think like a lawyer.” Clients have always and will always value this expertise. In the past, clients hired top-flight law firms because they needed assurance that whatever problems came up; whatever novel, complicated, nuanced legal issues arose, the firm would have the necessary expertise to handle it. But in actual practice, those complicated questions are rare, and only a tiny fraction of the billable hours needed to handle any given litigation matter represent the actual exercise an attorney’s expertise.

In most other industries, such conditions would quickly lead to the more repetitious and less cognitively demanding tasks in any given project being assigned to technicians, while experts performed higher-level strategic planning. But attorneys are sworn to uphold standards of professional ethics and conduct, and observing those standards and fulfilling those duties means that attorneys cannot or should not entrust tasks that may impact the quality of their client’s representation to non-experts, regardless of how menial they might be.

These professional standards are necessary to maintain the integrity of the legal profession, but frankly, it’s wasteful to employ legal experts as legal technicians. The negative consequences of this are twofold: First, it underutilizes the attorney’s most valuable asset: legal judgment and expertise. Second, it requires skills that new attorneys were never taught in law school, skills that law firms consequently must spend time and money teaching their new hires.

Let’s return once more to the fictional attorney and her AI assistant, and examine the division of labor between the two. The AI provides an overview of the relevant law and some tailored research leads. This eliminates the time the attorney might need to spend familiarizing herself with the foundational principles of an unfamiliar area of law and provides a starting point. But the nuanced task of reading individual cases and evaluating their precedential value is still done by the attorney; she just gets to it faster. The AI estimates the potential award or settlement amount for the case and provides the statistical basis for that estimate, but the attorney has the responsibility of interpreting this estimate and reconciling it with the client’s needs and desires. The AI can draft the initial set of routine litigation documents, but the attorney must still exercise her judgment when she decides whether or not those documents are ready to file (they typically are nearly 90 percent file-ready). In each of these instances, the portion of the task that requires an attorney’s expertise and judgment is reserved to the human attorney.

Such dramatic changes from the status quo bring risks, of course. The main risk is that the attorney will become overly reliant on technological aids and stop exercising legal judgment. An attorney who, out of ignorance or laziness, takes the summaries of law provided by artificial intelligence tools as the gospel truth, or who fails to review the machine-generated documents before signing off on them, would be in clear violation of ethical duties and could potentially be guilty of malpractice. Some will be tempted to just trust the AI because it’s correct more often than not, and professional responsibility courses will need to address and warn against this temptation. However, the benefits that legal professionals and their clients stand to reap from AI far outweigh these risks.

From a business perspective, each individual lawyer in a firm becomes more productive. This affords firms that integrate AI into their practice a serious competitive advantage, and empowers them to offer their clients more value for less money. There are substantial benefits for the individual attorney as well, especially for fresh graduates: A law firm’s new associates will require less training and spend less time performing routine tasks. This frees them to perform more cognitively demanding tasks and exercise their legal expertise. Associates will have the opportunity to do intellectually challenging work more often, gaining experience and maturing as attorneys faster than before. These attorneys will find the early parts of their career more rewarding, will in turn provide more value to their firms and their clients.

About the Author

Thomas Suh is the co-founder and president of LegalMation, an AI-based legal technology company that produces draft litigation documents. He is also the managing partner of LTL Attorneys LLP, a complex business litigation boutique based in California. He can be reached at

Send this to a friend