GPT-3 and its Impact on Ethics in the Practice of Law

Megan McCullough* and Amy Cyphert

College of Law, West Virginia University, Morgantown, WV 26506

Presentation Category: Oral-Human Engagement (Oral Presentation #9)

Student’s Major: Forensic Chemistry

The introduction of Artificial Intelligence tools into the practice of law has raised issues of ethics and raised questions about the adequacy of the Model Rules of Professional Conduct (a continuously updated set of ethical rules and professional responsibilities for lawyers created by the American Bar Association that is used in most states). GPT-3, the third generative pre-training model created by OpenAI, has made its way into the hands of beta testers and will soon be commercially available to everyone, including lawyers. This specific software is a natural language processor that recognizes writing pattern. It uses these patterns to generate writing, and presumably it can draft legal documents after receiving an input of just a few sentences. GPT-3 is better than many other AI programs because it can produce documents faster than a human and be refined on the users’ documents. GPT-3 has weaknesses such as bias and was trained on a dataset that included sources, such as Redditt, which can question the accuracy and usefulness of its output. Research is being conducted by reading articles of the basic structure and functions of GPT-3 and its reviews on the demos from technology professionals. There are strengths in using GPT-3 in the practice of law when it comes to efficiency, but the issue of ethics with this software raises a major concern in using it in the justice system. This specific AI technology may not have been used by lawyers yet, but there are potential violations to rules in the field.

Funding:

Program/mechanism supporting research/creative efforts: WVU's Research Apprenticeship Program (RAP) & accompanying HONR 297-level course