Generative AI chatbots like ChatGPT and Microsoft Copilot are powerful, user-friendly tools that can assist with a variety of tasks and activities like writing, research, and creative work. However, the types of responses you receive can vary significantly based on the messages you write (known as "prompts") and how you interact with them.

This guide offers practical tips for crafting better prompts and engaging with these tools more effectively to improve the quality, relevance, and efficiency of their output.

1) Start with a clear objective

Before writing your prompt, clarify your goal. Are you looking to generate a discussion question for a class, find resources on a specific topic, or brainstorm ideas for a group project? A clear objective helps with coming up with precise prompts that yield relevant and useful responses.

2) Be specific and detailed

• Avoid generic prompts: Rather than asking broadly about a topic, zoom in on the specific aspects you're interested in. For example, instead of "Tell me about the history of the Internet," try a more focused prompt, such as, "Provide an overview of the Internet's influence on pop culture, highlighting key figures and events."
• Adjust for tone and target audience: Clearly define the tone and who your intended audience is. If you're creating content for an educational setting, specify that it's intended for students or a particular course level. An example might be, "Draft an introductory lesson plan on machine learning algorithms tailored for senior computer science majors, designed to be engaging and technically detailed."
• Define your desired format and length: Specify the structure and size of the expected response. This helps in getting output that matches your exact needs, whether it's a brief summary or a detailed report. For example, you could request, "Write a five bullet point summary outlining the basics of continental philosophy."
• Incorporate context when necessary: Provide context about your situation or needs. If you have specific constraints or goals, mention these in your prompt. For example, "Suggest some Python beginner projects that can be completed in a weekend, focusing on data analysis."
• Use "do" or "do not": Mention elements that should or should not be part of the chatbot's response. For instance, "Please include key statistics but do not delve into technical jargon."
• Share examples or drafts for guidance: If you have an initial draft or examples that capture the essence or style you're aiming for, include them in your prompt. This approach helps the chatbot understand your intention more clearly. An example prompt could be, "I'm working on a lecture about urban sustainability. Here's my draft so far; can you continue in this tone?"

3) Review and refine

• Build and iterate on previous prompts: Follow up, refine, and ask questions after your initial prompt. You can start with a foundational prompt and then enhance it through subsequent interactions. For example, "Based on your previous explanation, can you now focus on case studies that illustrate successful initiatives?"
• Ask why it generated the response it did: Understanding the chatbot's reasoning can expand your perspective and offer room for further discussion or adjustment. For instance, "Can you explain why you chose this particular example?"
• Provide corrections and feedback: Address any inaccuracies, misunderstandings, missing content, or other undesired output. This helps tailor the chatbot's future responses more closely to your expectations. An example might be, "The information on renewable energy was useful, but can you include more recent statistics?"
• Start a new chat: Sometimes, despite refining prompts, the dialogue might not progress as you'd hoped. Starting a new chat can help reset the context and allow you to approach the topic from a different angle or with a fresh set of instructions. This can be especially helpful if the conversation has become too complex or has veered off topic.

4) Experiment with different techniques and approaches

• Use open-ended questions to brainstorm: Open-ended questions can help when you're in the brainstorming phase or looking for a wide range of ideas. For example, asking "What are some innovative ways to teach coding to young adults?" can generate a variety of suggestions and perspectives that you might not have considered.
• Role-play: This technique is useful for generating specific types of content or simulating conversations. You can ask the chatbot to assume the role of an expert in a field, a character from a story, or even a historical figure to provide insights or responses from that viewpoint. For instance, start your prompt with something like, "You are a nutrition expert advising on a plant-based diet for athletes."