What dataset is Spellbook trained on?
Spellbook is built on OpenAI's GPT-4 and other large language models (LLMs). It has been trained on a massive dataset of 42 terabytes of text from the Internet as a whole, contracts, books and Wikipedia. Spellbook is further tuning the model using proprietary legal datasets.
Dec. 15, 2023
Is Spellbook aiming to give advice to non-lawyers?
No. Spellbook is strictly available for legal professionals today. It is a tool to help professionals work much more efficiently, very similar to GitHub Copilot for programming!
Dec. 15, 2023
What if Spellbook gives bad legal advice?
Spellbook's output should always be reviewed by a lawyer. It is not meant to provide advice to end-clients or to replace human judgement. It is simply a tool that helps lawyers work more efficiently.
Dec. 15, 2023
Has Spellbook been tested for bias?
The OpenAI and Spellbook teams are working on an ongoing basis to detect and mitigate bias in their products. While AI will never be perfect, we believe it can ultimately be trained to be less biased than the average human.
Dec. 15, 2023