Machine learning researchers and legal experts have released SauLM-7B, which they claim is the first open-source text-generating large language model specifically focused on legal work and applications.
How large language models work
In light of recent high-profile mistakes in which generative AI cited non-existent cases in court filings — Mata v. Avianca and Park v. Kim — that may seem inappropriate. The tendency of AI models to hallucinate and their uncertain data provenance appears to be business in an industry where the stakes are significant.
But the creators of SauLM-7B, affiliated with the startup Equall.ai, Université Paris-Saclay and Sorbonne Université in France, and Universidade de Lisboa and NOVA School of Law in Portugal, argue that there is a place for artificial intelligence assistance in the law.
"LLMs and more broadly AI systems will have a transformative impact on law that includes but goes beyond marginal productivity," an Equall.ai spokesperson said in an email to The Register. “Our focus is on creating end-to-end legal AI systems that are governed and controlled by lawyers.
