1. Using Mnemonic Devices for Better Prompting
Using mnemonic devices like ICE (Intent, Context, Example), RISEN (Role, Instructions, Steps, End Goal, Narrowing), or COSTAR (Context, Objective, Structure, Tone, Audience, Result) can help you craft effective prompts. They remind you to include essential elements, making your prompts clear and comprehensive:
ICE:
- Intent: Clearly state what you want to achieve.
- Context: Provide background information to set the stage.
- Example: Give an example to guide the AI.
RISEN:
- Role: Assign a role to the AI.
- Instructions: Be specific in your instructions.
- Steps: Outline the steps to follow.
- End Goal: State the desired outcome.
- Narrowing: Provide constraints to focus the response.
COSTAR:
- Context: Provide the background information.
- Objective: Specify what you want to achieve.
- Structure: Define how the information should be presented.
- Tone: Dictate the manner the response should be delivered.
- Audience: Specify who the response is intended for.
- Result: State the expected result.
These frameworks help ensure your prompts are clear, comprehensive, and effective, leading to more accurate and relevant responses from #Copilot and other GenAI tools on the first try, reducing the need for multiple iterations.
2. Zero-shot vs. Prompt-stacking vs. Prompt-chaining
There are many prompting techniques you can use to improve the responses from #Copilot and other GenAI tools—these are the three I use daily. Knowing when to use zero-shot, prompt-stacking, or prompt-chaining will absolutely enhance your AI interactions.
Zero-shot involves giving the AI a task without any prior examples, ideal for straightforward queries.
Prompt-stacking combines multiple prompts into one to refine the AI’s response, useful for complex tasks. Prompt-chaining links several prompts in a sequence to handle complex tasks in stages. Each prompt builds on the output of the previous one, allowing for a more detailed and refined response. This method is particularly useful for larger tasks that require step-by-step instructions or multi-layered analysis. For example:
- Step 1 Prompt: “Summarize the key facts of the case Smith v. Texas using bullet points.”
- Step 2 Prompt (using Step 1 output): “Identify the main legal issues presented in the case Smith v. Texas.”
- Step 3 Prompt: “Provide a detailed analysis of the court’s decision in Smith v. Texas, including the reasoning behind the ruling.”
- Step 4 Prompt: “Draft a conclusion summarizing the implications of the court’s decision in Smith v. Texas for future cases.”
The Benefits of Prompt Chaining
- Improved Accuracy: By breaking down a complex task into smaller, manageable steps, prompt chaining helps ensure that each part of the task is addressed accurately. This reduces the risk of errors and enhances the overall quality of the response.
- Enhanced Clarity: Prompt chaining allows for a more structured approach to problem solving. Each step can be clearly defined and executed, making it easier to follow the logic and reasoning behind the AI’s response.
- Flexibility: This technique provides the flexibility to adjust and refine each step based on the output of the previous one. If a particular step needs more detail or a different approach, it can be modified without affecting the entire process.
- Efficiency: By handling tasks in stages, prompt chaining can save time and effort. It allows for a more efficient workflow, as each step can be completed and validated before moving on to the next.
- Scalability: Prompt chaining is scalable and can be applied to a wide range of tasks, from simple queries to complex projects. It is particularly useful for tasks that require a high level of detail and precision.
Choosing the right prompt for the job is just another aspect of learning how to get the most from #Copilot and other GenAI tools.
3. Talk to AI Like a Human
When interacting with #Copilot and other GenAI tools, treat it like a conversation with a human, not a Google search. Be polite and clear in your requests.
Example:
Imagine you need to draft a non-disclosure agreement (NDA) for a new business partnership. Instead of typing “NDA template,” you could say, “Hi, I need to draft a non-disclosure agreement for a new business partnership. Can you help me create a template that includes clauses for confidentiality, duration, and mutual obligations? Thank you!”
By providing context and being polite, you make it easier for the AI to understand your specific requirements and deliver a more tailored response. This approach not only improves the accuracy of the output but also makes the interaction more pleasant and efficient.
Statistics on Politeness: Research has shown that being polite to AI can lead to better responses. A study conducted by Future Publishing revealed that more than 55% of respondents consistently use polite language when interacting with AI, up from 49% in the previous survey[1].
Additionally, a report from TechRadar highlighted an experiment where polite interactions with ChatGPT resulted in higher quality responses[2]. The study found that politeness encourages the AI to provide more thoughtful and accurate answers, creating a positive feedback loop where users continue to be polite, leading to even better responses over time.
References
[1] Tom’s Guide: New survey reveals how polite we are to AI assistants ahead of new …
[2] TechRadar: The Impact of Politeness on ChatGPT Responses
4. Validate Information
Always validate the information provided by #Copilot and other GenAI tools. There have been instances where lawyers cited fake case law due to relying solely on AI-generated content. Remember, if it’s wrong to do something without AI, it’s wrong to do it with AI. Double check facts and sources to maintain your credibility and avoid potential pitfalls.
Examples of AI-Generated Errors:
- Mata v. Avianca, Inc. (2023): Lawyers cited six non-existent cases generated by ChatGPT. The court fined them $5,000 and required them to send letters to six judges who were falsely identified as the authors of the fake opinions[1].
- Monk v. Goodyear Tire & Rubber Co. (2024): Lawyer cited two non-existent cases and multiple fake quotations. He was sanctioned $2,000 and required to take Continuing Legal Education (CLE) on AI[1].
- Wadsworth v. Walmart (2025): Lawyer cited eight non-existent cases. The fine is pending, the lawyer was replaced by his supervisor, and the firm paid opposing counsel’s fees and expenses for the court filing[1].
While the monetary sanctions might not be steep, the hit to one’s reputation is. Relying solely on AI generated content without verification has consequences. Always cross-check facts and sources to ensure accuracy and maintain credibility.
5. Use AI as a Content Accelerator
Think of #Copilot and other GenAI tools as a content accelerator, not a finished product generator. It can help you brainstorm ideas, draft content, and refine your writing, but always review and edit the output. This ensures the final product aligns with your voice and meets your standards.
Benefits of Using AI as a Content Accelerator:
- Idea Generation: AI can help you come up with new ideas and perspectives that you might not have considered.
- Drafting Content: AI can assist in drafting initial versions of your content to provide a solid foundation that you can build upon, saving you time and effort.
- Refining Writing: AI can help you refine and polish your writing. You can use it to check for grammar and spelling errors, improve sentence structure, and enhance the overall readability of your content.
- Consistency: AI can help maintain consistency in your writing style and tone (particularly useful for large projects or collaborative efforts) to standardize the language and format for a cohesive final product.
- Efficiency: Using AI as a content accelerator can significantly increase your productivity so you can focus on higher-level tasks.
By using #Copilot and other GenAI tools as a content accelerator, you can streamline the content creation process, enhance the quality of your output, and ultimately deliver more value to your clients and your organization.
6. Avoid the "AI Accent"
The “AI Accent” refers to the telltale signs of AI-generated content, such as overly formal language, repetitive phrases, or unnatural sentence structures. To avoid this, personalize and humanize the output by adding your unique touch. This makes the content more engaging and relatable to your audience.
Tips to Avoid the “AI Accent”:
- Review and Edit: Always review and edit AI-generated content to ensure the content flows naturally and sounds human.
- Add Personal Touch: Incorporate your unique voice and style into the content. Add anecdotes, examples, or personal insights to make it more engaging and relatable. 3. Use Natural Language: Write like you talk. Use contractions, casual language, and conversational phrases for authenticity.
- Vary Sentence Length: Mix short and long sentences to create a natural rhythm, keep the reader engaged, and make the content feel more dynamic.
- Read Aloud: Read the content aloud to check for awkward phrasing or unnatural sentence structures to ensure the content sounds natural.
We hope these tips help you connect better with your audience while accelerating your work.