SauLM-7B is based on the open source Mistral 7B model and works by responding to questions or prompts in natural language, focusing on law and legal issues. The creators believe that the inaccuracies of AI models can be mitigated and that LLMs specifically trained on legal data will hallucinate much less on legal topics than their generalist counterparts. However, they caution that AI models should not be relied on as if they're a legal database, and that double-checking the output of LLMs is advised.
Key takeaways:
- 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.
- The creators of SauLM-7B, affiliated with startup Equall.ai and several universities, argue that AI systems will have a transformative impact on the practice of law, providing greater precision and more useful tools for lawyers.
- Equall.ai believes that almost all legal work can benefit from AI, and that the inaccuracies of AI models can be mitigated, especially when they are specifically trained on legal data.
- Jonathan Schwarz, co-founder and chief scientist at UK-based legal AI startup Safe Sign Technologies, praised the makers of SauLM-7B for their sensible approach to specializing general LLMs, but noted that there's still work to be done.