From Casebooks to Code
How Legal AI Will Change Legal Education for the Next Generation of Lawyers
For all of us who have attended law school, not much has changed in the last 50 years. We take various fundamental classes, property, corporations, evidence, etc and review precedent setting cases. We learn to apply facts to the law and evaluate judicial decision-making in context. There are not 100 categories of information lawyers need to integrate. There are essentially three: Statutes/Regulations, Case Law, Facts.
Two of the three of those are not suitable for memorization and are readily accessible through free or low cost resources. The final category changes with every case a lawyer handles. How might AI affect legal education now and in the future.
ChatGPT Is Not the End. It Is Just the Beginning.
So much about AI currently is the wide adoption of ChatGPT by many non-techies. It is not the only tool that will spur education, but is symbolic of the future of education widely, just not post undergrad programs like law school.
Tools like ChatGPT, for example, can provide instant feedback and answer questions in real-time, allowing law students to learn at their own pace and gain a deeper understanding of complex legal concepts. The LLMs (AI models) like what ChatGPT was trained on can provide students with access to vast amounts of legal information and research, allowing them to explore topics in depth and gain a more comprehensive understanding of the law.
Those models can analyze legal texts and provide insights into legal language, helping students to better understand the nuances and complexities of legal writing. This can help students to develop better legal writing skills and become more effective advocates. Many of these tools going forward will be used by students and lawyers alike to improve their brief writing and, eventually, write entire briefs on their own.
I remember my days in law school which often involved trying to determine what answer the professor wanted to read in a test response. Law professors using AI models can automate the grading of class assignments and assessments, saving them time and providing more objective and consistent student feedback. Legal AI tools can augment instruction by identifying areas where students may need additional support or instruction.
AI models can be used to simulate legal scenarios, allowing students to gain hands-on experience and develop practical negotiating and courtroom skills. This would help prepare students for real-world legal situations and enhance their professional development far beyond just memorizing cases and briefing legal issues.
The online learning platform Khan Academy is already using AI tutors to expand its reach to millions of online learning students. Legal AI models could help increase access to legal education by providing online learning platforms accessible to students from anywhere in the world. This would democratize legal education and provide opportunities for students who may not have had access to traditional law school programs. The downside for law schools is that this ubiquity of access might well reduce the tuition they can charge students.
The announcement of GPT-4 on March 14, 2023 contained content on the webpage that went right at the ability of the LLM (large language model) to successfully answer questions required to pass the bar exam.
As mentioned above, Khan Academy, is already using GPT-4 as an AI tutor for students.
Harvey.ai is still in waitlist mode. But it is also featured on the GPT-4 website as having already adopted this newest LLM. Harvey summarizes its features as:
Harvey AI assists with contract analysis, due diligence, litigation, and regulatory compliance and can help generate insights, recommendations, and predictions based on data. As a result, lawyers can deliver faster and more cost-effective solutions for client issues.
OpenAI has been touting the dramatic increase in its model’s scores on various academic assessment and entrance exams. This graphic shows the score that the previous version GPT3.5 achieved on these various exams (the blue bars) and how much higher GPT 4 did on those same exams. In less than six months, the GPT LLM will be the highest score on all of these exams. Source
The true equity move here for law schools is to provide wider access to legal education. However, to do so would cause a major restructuring of law school curriculum and perhaps even the requirement of a bachelor’s degree prior to admission. We have all been in the presence of people with sharp minds, analytical skills and a passion for argumentation who never stepped foot on a college campus. Would such people enhance the legal field if they could just get access to legal education and AI tutors along the way? There seems to be a strong argument that the answer is yes.
Education has not yet been severely disrupted by technology for one main reason: The so-called Bloom’s Two Sigma Problem. Educational psychologist Benjamin Bloom reported a comparison between tutored students’ performance and the performance of students in a class where the teacher to student ratio was 30 to 1. His study indicated that tutored students, regardless of their previous educational level, performed better than 98% of the students in the regular classroom setting. So, at the time of Bloom’s discoveries, tutoring each student is the superior educational approach. But, in private and public education, a tutor for every student is cost-prohibitive. But, what about now when it isn’t?
AI is already being implemented in a tutoring role as noted above. We are now months away from tutors exploding on the scene in every conceivable area of education. It is not years away. Just consider having a personal tutor on your computer or phone and you can ask it any question and it responds and then poses new questions to you in an entertaining way, links to videos or podcasts about the content continuously prompting you into your areas of interest and exposing you to new areas. The quality and speed by which people learn is going to grow exponentially.
What does this mean for law professors? AIs are going to quickly be superior to human instructors at the learning part of education. They won’t be as useful at the content creation part. Therefore, the best of the best law professors will be those that can create the most engaging content.
Like all education, LLMs are going to transform legal education and higher education as well. Since these models will only continue to learn and never forget what they learned in the past, their utility as tutors in place of professors is going to be apparent. Online learning platforms are already embracing AI enhancements and law schools will have a choice - ignore the evolution and have the wave carry it wherever it leads or jump into the Legal AI environment and have some ability to control how it changes legal education now and in the future.