“Lawyers who use AI will replace lawyers who don’t.” That was the coda of a presentation given at the recent HPC User Forum in Tucson by Arizona State University law professor Gary Marchant, a graduate of Harvard Law School and a professor at ASU since 1999.
Marchant’s presentation at the Forum, hosted by Hyperion Research, was one of the conference’s most talked-about sessions. In this interview, Marchant shares his observations on how generative AI and large language models are profoundly impacting the legal industry and how services are delivered and billed.
Marchant cited numerous examples in which, for example, a lawyer preparing a privacy policy using ChatGPT, which enabled him to complete in 15 minutes what normally required six or seven hours. The lawyer faced a quandary: should he bill the client for six or seven hours of legal services, or 15 minutes?
Another impact: Marchant said AI is eliminating much lower-level legal work performed by recent law school graduates. This means newly hired lawyers will be expected to perform from the start at a higher level. It also means there will be fewer lawyers.
AI is also impacting case law itself. Marchant discussed legal liability for a case in which a factory robot struck and killed an assembly line worker, and another case in which a large language model “hallucinated” a sexual abuse case against a law professor. Who’s liable in such cases in which there was no human intent?
Finally, Marchant has suggestions for firms starting their AI journeys. Should they hire a data scientist and build a custom LLM? Or should they begin at a more modest and less costly level?
