Legalcurrent.com

Terius Madrigal

For every student or graduate looking to potentially go to law school in the future, this article is for you. It’s always important to know what future job opportunities look like for the career you plan on pursuing. As I plan to go to law school myself, I did an analysis on current and future trends regarding employment for lawyers. Currently there are 707,160 lawyers employed nationwide, with a median annual wage of ​135,740 dollars. Future employment looks positive, with the Bureau reporting that from 2022-2032 there will be an 8% job growth in the legal market. When it comes to potential changes that can affect lawyers in the labor market, there is one in particular that stands out: Artificial Intelligence (AI).

AI In the Legal Labor Market

Media sources state the complementary nature AI will have in the legal field. According to market research analysts, if lawyers become proficient in using AI software then firms can get ahead by feeding them case files and documents in order to improve efficiency in legal processes. They can have AI summarize an entire multi-page brief, search for keywords or statements, conduct a quick online search for past precedent that may apply to the case, etc. On the surface it seems like AI will have nothing but a positive effect on lawyers. However, when taking into consideration the complex nature of legal employment, incorporation of AI is not as straightforward.

Theoretical Effects of AI on Legal Employment

AI may reduce the need for law clerks. As law clerks are the ones who conduct legal research, analyze cases, and prepare legal documents, use of AI will make their lives much easier and more efficient. However, the efficiency of AI would decrease the marginal utility of the demand to hire an additional law clerk, since AI performs work that once took a worker days in a matter of minutes. Therefore a firm will have a lower threshold in the amount of workers they hire before they gain no added benefit from an additional law clerk.

A reduction in employment in law clerks is concerning for various reasons. The first is that law clerks are typically recent law school graduates, and it is one of the only jobs available straight out of law school in many firms. This means diminishing employment turnout for law school graduates, leading to less lawyers climbing the legal ladder, and less experienced lawyers in the market.

However, this outcome is the worst case scenario which does not factor many alternate turnouts. For instance, the efficiency of using AI for case research and analysis may allow cases to be dealt with so quickly that it is actually unprofitable not to hire more lawyers to take on cases. If AI assistance allows one employee to do the job of three, they may also be given over three times the workload they previously had. Therefore, the firm would actually take on more cases, leading to greater benefit from hiring more workers. This could potentially lead to a higher level of employment if the new marginal benefit per case outweighs the marginal cost per worker at a greater level than previously without AI assistance.

It also must be noted that AI can and will make mistakes. When it comes to interpretation of the law there are high stakes at play. Mistakes can’t be afforded as it results in permanent consequences to individuals, firms, or even the country as a whole. Thus, even if increasing reliance on AI would decrease the need to hire graduating lawyers for research, it would increase the need to keep or hire a lawyer to double check the AI’s findings to ensure there isn’t an error. 

Even if firms can take on more cases per lawyer there is also projected increased demand for lawyers, so it is unlikely that supply of lawyers will outweigh the demand. This is because becoming a lawyer is not only expensive but also difficult. One must complete an undergraduate degree to pursue law school, as well as take on the financial strain that is attending law school. It’s too big of an investment for a rapid increase of supply to suddenly occur due to greater demand.

An Empirical Study Measuring Performance of Legal Tasks with AI

As in other fields, AI will likely be integrated into students’ legal curriculum. To see AI effects in the classroom unfold as well as analyze how AI use may impact future legal employment levels, we observe the results of a study conducted by Jonathan H. Choi and Daniel Schwarcz. Their study observed law students’ performance on exams with and without the help of AI. The primary objective of this study was to observe changes in performance in lawyers with AI, and whether such changes are likely to cause elimination of legal jobs in the market.

The AI used in this study was GPT-4. Students’ results were divided into two groups. The experimental group were lawyers with access to GPT-4 on a second set of final exams. Meanwhile the control group were students during the first set of final exams, where they did not have access to GPT-4. Prior to taking the second set of final exams students were given thorough training on how to effectively use GPT-4 for their exam and studied material.

They found that AI assistance had mixed results on exam performance. On multiple choice questions student’s using AI improved by 29% compared to those without access. However, there were no apparent differences in grades on the essay portion of exams. They concluded that for more complex aspects of the law AI is likely to have little effect. It was also found that a student’s improvement on the exam due to AI significantly depended on the student’s performance without it. Poor performing students significantly benefited from AI assistance, scoring 45% higher than their base results. Meanwhile, the best performing students actually scored lower after the assistance of AI, declining significantly at 20% lower.

Choi and Schwarcz also tested GPT-4’s performance alone on exam material, finding substantial variation. In both basic introductory legal classes and complex level classes AI outscored both humans who used AI and humans who didn’t. This raised concern over the potential for AI to replace humans in the legal field, particularly paralegals and law clerks. However, such results also raised the idea of increased time available for lawyers to focus on more cases and improved efficiency.

Employment Predictions Based on Study Findings

Based on the findings of Choi and Schwarcz we can confer a few trends. The first is that AI will likely even out the legal landscape, leading to on average increased performance amongst lawyers in resolving cases. With demand for lawyers and legal work going up, this improved AI backed efficiency leads to cases getting resolved quicker with less prone for error. We can also conclude that more complex work will likely not be affected by AI and will remain at roughly the same pace and efficiency as it does now. Therefore lower level legal services will see improvements and potential increased hiring levels if firms gain greater marginal benefit from AI assisted labor than marginal cost of hiring. Meanwhile those in upper level legal positions will likely not see a great amount of employment change as their skillset remains roughly the same regardless of AI assistance. This conclusion, however, only applies in the short run with current AI capabilities.

The long run effect of AI on legal employment may be more grim. With AI able to outperform both students who used AI and those who didn’t, it raises concerns for the future; particularly, one where AI reaches the capacity to perform tasks without the vast amount of inputs needed to direct it. If AI reaches a level where it can resolve basic cases entirely on its own, answer theoretical questions, and represent a client better than a human can, the legal field is in vast trouble. Even in a future where it still requires some inputs to direct it, it is likely lower level lawyers like clerks, and those without a legal degree like paralegals, that will receive reduced employment.

Long Run Legal Optimism 

As scary as this may sound, it is also still unlikely to occur, as there are relatively high barriers to entry to become a lawyer. You need at least a college degree to go to law school, and then must dedicate the proper time and effort to even become a lawyer. Not to mention most lawyers come from an educated family background, with 73% of law students in 2019-20 having a parent who at least completed college. Additionally, many top law firms only hire from schools they trust. As a result, unless one goes to a law school with a stellar reputation, it is difficult to find employment outside the regional area that schools are familiar with. There is a systematic setup when it comes to legal employment, one that is unlikely to be easily dismantled by technological innovation. All this to say that it is unlikely law firms will hire an AI to do the job rather than a lawyer fresh out of law school. The legal field is a social one, where reputation and trust are factors aside from performance alone.

Additionally there is the client aspect of the legal field. Clients, who are paying a large sum, would likely prefer humans to represent them than AI. One wouldn’t want to be given an injection by a robot over a human nurse, as the nurse can understand and convey human emotions and pain to a greater extent. The same goes for law, clients want a worker who can understand their emotions and desires. A lawyer can convince a jury by the way they frame a case and evidence, rather than through the evidence alone. This is an innate human trait, something that cannot for the foreseeable future and, potentially never, be replicated by an AI.

Conclusion

Ultimately, even if AI begins to make its way into the everyday lives of lawyers, it is unlikely to affect employment equilibrium significantly. The most likely case is that the demand and supply of lawyers will both increase steadily over the next decade and therefore equilibrium will be maintained. If you are a student looking to see if law school is still worth it 10 years down the road you are in luck. For the foreseeable future the legal field should continue to hold steady levels of employment growth. As AI progressively advances however, more variables may come into play.