AI BriefingBuilding a Custom Model for Legal Professionals
Key point
Harvey partnered with OpenAI to develop a custom model for legal professionals trained on vast amounts of case law data.
Details
Harvey partnered with OpenAI to build a custom case law model for legal professionals. The platform is a secure generative AI platform for professionals in law, tax, and finance, and recently raised $80 million in a Series B funding round, reaching a valuation of $715 million.
Existing RAG (Retrieval-Augmented Generation) or simple fine-tuning approaches had limitations in solving complex, open-ended legal research cases. To address this, Harvey collaborated with OpenAI to train the base model on about 10 billion tokens of case law data in order to inject new knowledge and reasoning methods into it.
Test results showed that 97% of lawyers at large law firms preferred Harvey's custom model over GPT-4. The model provides more detailed and complete answers, with every sentence backed by cited case law, effectively reducing hallucination.
Harvey plans to focus on using this model to develop AI agents that assist with drafting documents and legal review going forward.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.