AI-Prompt Engineering for Lawyers and Law Students: Tips to Get the Most Out of AI Tools
In today’s fast-paced legal landscape, leveraging Generative AI can significantly enhance your research, drafting, and case management. But getting the best results from AI tools like GC AI, Marveri, Responsiv, Alexi and Spellbook often depends on the quality of your prompts.
One Legal AI company that focuses on this is GC AI. Once you sign up for the 14-day free trial, Founder Cecelia Ziniti sends you a daily ‘drip’ email campaign with insightful prompting Tips & Tricks, and also teaches a prompting crash course on Maven. GC AI also ‘builds-in’ some novel prompting technologies that continue to delight users.
All “Off-the-shelf” and Custom Legal AI Solutions require some form of ‘prompting’ and conversation between a lawyer and the AI, and lawyers are getting better at prompting in part due to the nature of their training and skills.
Meantime, here's an abbreviated guide to prompt engineering that will help you harness AI's full potential in legal practice.
1. Be Specific and Direct
Vague prompts can lead to vague results. When crafting questions or commands for your AI tool, be as detailed as possible. For example, instead of asking, “What’s the law on breach of contract?”, refine the prompt to, “Provide key precedents on breach of contract cases in New York after 2020.” The more targeted your query, the better the response.
2. Use Legal Terminology Wisely
AI tools are trained on large datasets, but their understanding improves when you use clear legal terminology. If you’re looking for case summaries, cite the specific legal terms or doctrines you want to explore, such as "Res Ipsa Loquitur in personal injury cases". This precision helps the AI deliver relevant case law or statutory interpretation.
3. Break Complex Queries into Steps
For multi-part questions, separate them into smaller, manageable parts. Instead of asking, “What’s the legal precedent and its application in breach of fiduciary duty in the last five years?”, split it into two:
“What are the leading precedents for breach of fiduciary duty?”
“How has this precedent been applied in cases from 2019 onward?”
This method helps the AI provide more coherent and comprehensive answers.