Why It Matters
Fine-tuning can improve style and accuracy on specialist tasks, but it is expensive and the model has to be retrained as base models improve. Many legal vendors rely on RAG and prompting instead. If a vendor fine-tunes, ask whose data it was trained on.
Related Terms
- Foundation model
- A large, general-purpose AI model trained on broad data that other products are built on top of. Most LLMs are foundation models.
- LLM (large language model)
- The type of AI model behind ChatGPT, Claude, Gemini and most legal AI assistants, trained on very large amounts of text to predict and generate language.
- Training on customer data
- Whether a vendor uses the documents and prompts you send to improve its AI models.
General information, not legal advice. Part of the Legal Tech Glossary, 173 terms in plain English.