Understanding Prompts

Here’s a helpful video by our friends at GitHub about prompting.

Prompts are the instructions you give to UT Spark. Whether you’re asking it to summarize an article, draft a message, learn about a topic, or support an academic task, the quality of your prompt directly affects the quality of the response.

What Is a Prompt?

A prompt is any text input you send to UT Spark. It can be a simple request:

"Summarize this article."

Or something more specific:

"Provide a 250-word summary of this article, highlighting the main arguments and supporting evidence."

Both are valid, but the second example gives the model more structure and usually results in a better response.

Components of a Good Prompt

• Clarity – State what you want as directly as possible. Avoid vague language.

• Context – Include relevant background if UT Spark needs to understand the situation, source material, or audience.

• Constraints – Set boundaries such as word count, tone, or formatting.

• Expected Output – Tell UT Spark what kind of response you want: a list, summary, email draft, outline, etc.

Best Practices for Writing Prompts

Be Specific

The more focused your request, the more useful the answer will be.

• ❌ "Explain photosynthesis."

• ✅ "Explain photosynthesis in 3–4 sentences using simple language suitable for high school students."

Provide Context

If your request depends on background information, include it.

• ❌ "Summarize this."

• ✅ "Summarize this research paper in 150 words, focusing on its main hypothesis and conclusions."

Define the Format

Tell UT Spark what kind of output you’re looking for: list, paragraph, bullet points, step-by-step guide, etc.

• ❌ "Give me tips on writing a resume."

• ✅ "List five resume-writing tips for recent college graduates applying for academic research roles."

Specify Tone and Style

This helps UT Spark adapt its voice for your audience.

• ❌ "Explain quantum mechanics."

• ✅ "Explain quantum mechanics in a conversational tone as if teaching a first-year undergraduate physics student."

Break Complex Tasks into Steps

If you’re asking for a multi-step output, say so up front.

• ❌ "Help me write a lesson plan."

• ✅ "Provide a step-by-step guide to creating a lesson plan for a university-level literature course on 19th-century poetry."

Examples by Role

For Faculty

• "Create a 45-minute lesson plan on the basics of machine learning for undergraduate students."

• "Draft a compelling introduction for a research grant proposal on renewable energy in urban settings."

• "Generate five multiple-choice questions on the Renaissance period for an art history exam."

For Staff

• "Draft a professional email inviting guest speakers to an academic conference on AI ethics."

• "Summarize the university’s new attendance policy in under 200 words for a faculty handbook."

For Students

• "Outline a 1,000-word essay on the impact of social media on political activism."

• "Summarize key takeaways from ‘The Great Gatsby’ in bullet points for a study guide."

• "List five peer-reviewed sources on climate change adaptation in coastal cities."

Troubleshooting Prompt Issues

Helpful Tips

PACE Framework

Designed to help you create effective prompts, ensuring you receive accurate and relevant responses. Follow these steps to craft your prompts: 

• Problem: Identify the issue you want to solve. Use phrases like "I need to...", "I have to...", "I need help with...", or "I am struggling to learn " to clearly state the problem. 

• Action: Define the task you want UT Spark to perform using a clear action verb. Examples include: 

  • Create

  • Generate

  • Brainstorm

  • Illustrate

  • Draw

  • Envision

  • Craft

  • Conceptualize 

• Context: Provide any additional information needed to give a relevant and complete response based on the tasks and actions required. 

• Expectations: Influence the response by presenting examples and expectations for the desired output.