Introducing SauLM-7B: The First Legal-Focused Text-Generating Open Source Language Model

Machine-learning researchers and legal experts have released SauLM-7B, the first text-generating open-source large language model specifically focused on legal work and applications. Despite recent high-profile blunders in which generative AI cited non-existent cases in court filings, the creators of SauLM-7B argue that there is a place for artificial intelligence in the legal industry.

Affiliated with startup Equall.ai and several universities in France and Portugal, the creators of SauLM-7B believe that AI systems specialized for the legal domain will outperform generalist ones. They argue that AI can assist with various legal tasks such as research, document review, analysis, summarization, and identifying key passages in documents.

Other organizations, such as Goldman Sachs, have estimated that a significant portion of work tasks in the legal profession could be automated by AI. Startups like Bench IQ, Harvey.ai, and Safe Sign Technologies see a market opportunity in AI assistance for legal work.

The creators of SauLM-7B acknowledge that AI models are probabilistic and may have inaccuracies. They emphasize the importance of double-checking the output of AI models and caution against relying on them as a legal database.

SauLM-7B, named after the TV series Better Call Saul, is based on the Mistral 7B model and has 7 billion parameters. The creators of SauLM-7B describe their work in a paper titled “SaulLM-7B: A pioneering Large Language Model for Law.” The model is available on the AI model community site HuggingFace.

Jonathan Schwarz, co-founder of Safe Sign Technologies, commended the creators of SauLM-7B for specializing general language models for legal applications. Safe Sign Technologies is working on its own legal LLM and aims to deploy it through partners in the near future.

Schwarz highlighted the importance of red-teaming models to ensure their accuracy and reliability. He expressed skepticism about the approach taken by some AI companies and emphasized the need for a focus on safety and robustness in AI systems, especially in sensitive areas like law and medicine.

Overall, the release of SauLM-7B represents a significant advancement in the field of legal AI and opens up new possibilities for using AI to enhance legal work and applications.