If you follow tech news, you probably know about Chat-GPT, an artificial intelligence (AI) program that has been writing college essays, designing buildings, and even crafting condolence emails after mass shootings. Just last week, we learned that this program "can now outperform most law school graduates on the bar exam, the grueling two-day test aspiring attorneys must pass to practice law in the United States." The test consists of multiple choice and essay questions, and Chat-GPT's score placed it "in the 90th percentile of actual test takers," which "is enough to be admitted to practice law in most states."

Perhaps in an effort to make lawyers feel better, the "National Conference of Bar Examiners, which designs the multiple choice section, said... that attorneys have unique skills gained through education and experience that AI cannot currently match." "Currently" being the operative word.

While I do not expect to lose my job in the immediate future, I can see areas of immigration law where AI would be able to assist attorneys and increase efficiency. And I can also imagine a future where AI takes over many, if not all, tasks performed by attorneys, Asylum Officers, and Immigration Judges.

In my particular specialty--asylum law--one obvious area where AI could assist involves defining the applicant's "particular social group" or PSG. In order to qualify for asylum, a person must demonstrate that they face persecution on account of race, religion, nationality, political opinion or PSG. The first four protected categories are relatively self-explanatory. However, there is much debate about many PSGs, and the law has changed (repeatedly) over time. In short, defining PSG in a case can be difficult. It is also important: If the decision-maker does not accept your proposed particular social group, you will not be granted asylum.

The problem is that different courts have reached different conclusions about the same proposed PSG (see, for example, the controversy surrounding the PSG "married women in Guatemala who cannot leave their relationship"). As a result, there are all sorts of published and unpublished cases opining about PSGs. These include cases from Immigration Judges, the Board of Immigration Appeals, the federal circuit courts, and the U.S. Supreme Court. There are also various regulations, memos, and other documents from the government. In short, there is a lot of information out there, and it is not always easy to locate. This seems like a perfect job for AI--to gather and sift through tons of information and then use that to create a legal argument supporting the viability of a particular PSG.

Another area where AI could come in handy is defining "persecution." In order to win asylum, you need to show that the harm you face is severe enough to meet the definition of "persecution" under the law. Also, where a person suffered past harm in the home country and can demonstrate that that harm rises to the level of persecution, they can receive asylum in the U.S. (unless country conditions have improved). As with PSG, the problem here is that there are wildly inconsistent interpretations of what types of harm constitute persecution. And so this seems like another area where AI could synthesize the various decisions and reach a conclusion about whether the past harm or feared future harm equates to persecution.

AI could also help ensure that all affidavits and documents are consistent. It could compare each document to every other document, and make certain that all dates and events match up. Where there are inconsistencies, the decision-maker might think that the applicant is lying, and so checking for consistency (which is susceptible to human error) is important. AI could also check that the evidence and affidavits are consistent with country conditions, and find news articles or human rights reports that support the applicant's claim.

I could also see AI coming in handy for analyzing the effect of criminal conduct on an immigration case. This is a complicated area of law (dubbed "crimmigration"), which involves determining how state, federal or even foreign criminal law interacts with U.S. immigration law. I imagine AI would be able to gather and analyze this data to determine how a particular crime affects an immigrant's status in the U.S.

These are just a few examples of how AI could be applied to immigration law, and could help attorneys make more powerful arguments for our clients.

AI would also be helpful to Asylum Officers and Immigration Judges, as it will allow them to reach decisions that are more well-reasoned, more consistent, and more objective. AI could also be used to scan large numbers of cases to determine which cases might be adjudicated quickly--such a feature could allow the government to better prioritize cases and help reduce the backlog.

While AI could likely assist or take-over certain lawyerly and judicial tasks, other aspects of legal practice seem less amenable to machine assistance. For example, some areas where human power might still be superior include talking empathetically to clients in order to help them tell their stories, providing reassurance to people in a difficult situation, preparing witnesses to testify, and working with applicants who may be unable or unwilling to prepare a case on a computer.

Whether AI will ultimately play a supporting or a leading role in the practice of immigration law, I do not know. But given the rapid advances we have witnessed in recent months, I expect that AI will become more important and more common in our field. Hopefully, this will lead to better representation for applicants and better decisions from adjudicators.

Originally posted on the Asylumist: www.Asylumist.com