Improve Your AI Results by Writing Better Prompts

Questions

  • What is prompt engineering and why is it important?
  • How can I write effective prompts to get better results from AI tools?
  • Who can use prompt engineering at Bowdoin College?
  • Are there tips or examples that help improve AI-generated responses?
  • Where can I learn more about prompt engineering best practices?

Environment

This article is for Bowdoin College students, faculty, and staff who use AI-powered tools or chatbots, such as generative AI writing assistants. It applies to anyone looking to improve the way they interact with AI systems on Bowdoin-supported computers or online services.

Fun Fact: Bowdoin Bot, the chatbot found across the Support Portal, is a generative AI bot. Your results will benefit from using some of the recommendations found on this page.

Resolution

  1. Understand what prompt engineering is. Prompt engineering is the process of crafting clear and specific instructions, called "prompts," for AI systems. The goal is to get more accurate or useful answers from the AI by phrasing your request thoughtfully.
  2. Be clear and direct in your prompts. Phrase your questions or requests in a straightforward way. For example, instead of saying "Explain photosynthesis," you could say "Explain photosynthesis in simple terms for a high school student."
  3. Provide relevant details. Including important context, such as your audience or what format you want the answer in (a summary, a list, etc.), helps the AI give you a more tailored response.
  4. Set Expectations: If you want a specific format (like a bulleted list or summary), or a particular tone (such as formal or friendly), mention these in your prompt. Example: “Summarize the article in three short bullet points.”
  5. Ask for follow-up questions: Many AI models are capable of asking questions to get more information from you to provide the best answer. Don't be afraid to ask the AI to ask you follow-up questions to get a better answer.
  6. Break down complex requests. If you have a big or multi-part question, split it into smaller pieces or ask for step-by-step guidance. For example, "List the main causes of climate change, then explain one in detail."
  7. Review and refine your prompt if needed. If the first answer isn't what you expected, try restating your request more clearly or adding information. Experiment with phrasing to see what works best.
  8. Keep learning about best practices. For more information, visit Best Practices for Prompt Engineering by OpenAI or review this external guide on effective prompt writing: Prompt Engineering with Microsoft Azure OpenAI.

Additional Help

If you need further assistance, you can chat with Bowdoin Bot of contact the Bowdoin College Service Desk by phone at (207) 725-3030 or in person at the Tech Hub in Smith Union.