ESSENTIAL TECH

AI MODELS OUTPERFORM LAW STUDENTS IN KEY EXAMS, RAISING QUESTIONS FOR LEGAL TRAINING
August 14, 2025
By Adediran Peter A.
Artificial intelligence is no longer just a research tool — it’s now competing with top law students.
A recent University of Maryland study found that Open AI’s new “o3” reasoning model scored A+ grades in Constitutional Law, Property, and Professional Responsibility on real law-school final exams. The model’s ability to plan, revise, and self-correct gave it a clear edge over earlier AI systems.
In Australia, researchers tested AI on a Criminal Law exam and discovered that performance jumped from barely passing to top-quartile scores simply by improving the prompts. The finding highlights that how you ask AI can be as important as the AI itself.
Experts say the results could reshape junior legal roles, shifting focus from rote memorization to skills like strategic thinking, client engagement, and quality-checking AI outputs.
However, risks remain. AI can be out-of-date, make factual errors, or introduce bias. Legal educators and firms are being urged to pair AI tools with human oversight, strict verification, and ethical safeguards.
“AI won’t replace lawyers, but lawyers who use AI effectively will replace those who don’t,” one legal tech analyst told reporters.
"This represents a significant development in our ongoing coverage of current events."— Editorial Board