[行业应用] Real-Time Engineering in the Legal Profession: Beginner Tips

Should lawyers learn prompt engineering? Becoming a prompt professional is not a prerequisite for using generative artificial intelligence (AI). But understanding the basics gives lawyers an advantage.
To unlock the benefits of prompt engineering, two things need attention. First, how models use AI to retrieve and combine information; second, how to write effective prompts.

By writing effective prompts using plain language and your subject-matter expertise, you will get more accurate results. After all, AI must be supervised by (human) lawyers to ensure accuracy.

Here is how to write effective prompts based on the logic behind them.

What is prompt engineering for lawyers?

Prompt engineering is the practice of formulating targeted queries to extract relevant results from AI tools used by lawyers or solicitors. A prompt is simply a sentence written in natural language that provides the tool with the required task and any additional information. The practice is iterative.

Prompt engineers test, evaluate, and optimize results by trying wording, terminology, and instructions. Prompt engineering practice includes ethical considerations, meaning reviewing results and repeating prompts to eliminate bias or misinformation.

A well-crafted prompt will help generate accurate responses tailored to the user's needs. It will help reduce inaccuracies or so-called hallucinations.

What should lawyers know about prompt engineering?

Generative AI is redefining how lawyers work. AI legal assistants will help speed up tasks such as legal research, drafting, and analysis. After the heavy lifting is handled, you will free up more time to communicate with clients, develop strategy, and add value to the business. But first, you must know how to prompt.

Lawyers should incorporate their legal expertise into the logic of prompt engineering to produce better results. But to gain a competitive advantage as a legal expert, you need to understand some prompt engineering basics. A basic understanding of how to structure prompts will help you obtain valuable results from vast databases, preserve legal nuance, streamline your work, and improve the accuracy and quality of your work.

How should lawyers write prompts for generative AI tools?
It is all about clear communication. The simplest way to craft a prompt is simply to think about how you would effectively introduce the situation to people.

What is the outline of the task?
What do they expect to be provided?
What guidance do they need?
Is their context clear?
Are there any caveats?
The more accurate the prompt, the more accurate the result.

5 tips for writing effective legal prompts

1. Use precise language
Express your prompt clearly. Keep sentence structure simple and concise. Avoid being overly formal, so drop the passive voice and write in the active voice. It helps to think about how you would explain something to someone in conversation. This also means being inclusive, so spell out abbreviations and avoid slang. See these helpful before-and-after examples to learn how to fine-tune the wording of a prompt.

2. Provide context
For more complex prompts, provide any contextual information that helps define your query more clearly and account for legal nuances. Think of the tool as an actor who needs to become familiar with a role. Start by giving the tool an introductory statement that sets the scene and explains the role it needs to play.

3. Be specific
Clearly state the task and the expected format of the output. Provide examples where possible. This will help prevent the technology from filling any gaps, which can lead to imprecise results.

4. Refine the prompt
To refine results, you need to iterate the prompt. Think of it as a two-way conversation. Expect some back-and-forth to reach a consensus. The iterative process will ensure more accurate results.

5. Be responsible
Monitor your results for bias and pay attention to ethical considerations. Responsible AI relies on human oversight of results, and this is where you as a legal professional can apply your critical training in the field. Do you want it to explain its logic to you or cite all the sources it used? Ask it to.

Related reading: How does Australia regulate AI? What should lawyers know.

5 examples of good and bad AI prompts

Reviewing these examples will show you how to use these tools effectively in your daily work. We use CoCounsel (part of Thomson Reuters) as an example of a prompt recipient:

Prompt 1: "I represent a global corporate group, Hypothetical Example Holdings, which is seeking to divest part of its business. We are now in a vendor due diligence process. How can CoCounsel help me?"

Why it works: It provides context and first consults the tool for strategic advice. By understanding how the tool can help, you can improve prompts on the topic with new insights.

Prompt 2: "Please review this contract to determine whether it complies with the following policies: (1) Any anti-assignment clause should be mutual and should permit assignment in a merger, acquisition, or other change of control. (2) The governing law should be New South Wales."

Why it works: It is specific. It specifies the task and jurisdiction for CoCounsel to review.

Prompt 3: "Can you help me draft a letter? Address the letter to Jane Doe, counsel for the seller under this contract, Example Tech. I represent the buyer, Example Holdings.

"The letter needs to explain that I am sending back an updated draft of the agreement and explain why. Please summarize, based on CoCounsel's contract policy compliance report above, the ways in which the current agreement conflicts with my client's policies."

Why it works: It provides context and detail.

Prompts 4 and 5: "List the top five contracts in this database," or "Give me a legal contract I can use for my multinational client."

Why they do not work: Neither example is specific enough to produce useful results. In addition, the expected output (e.g., "contract") is not sufficiently explained.