Discussion: Improving a Prompt

Please submit all graded work via Canvas. Participation requirements and grading details are provided in Canvas.

This week we explored how people communicate with generative AI systems through prompts. In practice, the first prompt does not always produce the result we want. Often the output improves when the prompt is revised to include clearer instructions, more context, or a specific output format.

This discussion asks you to revisit a prompt you previously used and examine whether applying prompt engineering strategies can improve the result.

Step 1: Identify a prompt that did not work well

Choose a prompt you used earlier in this course or in your everyday use of generative AI.

The prompt should be one that did not produce a useful or satisfying result.

Briefly describe:

  • what task you were trying to complete
  • what prompt you used
  • why the response was not helpful

Include the original prompt in your post.

You may also include a short excerpt or screenshot of the AI response if it helps illustrate the problem.

Step 2: Revise the prompt

Rewrite the prompt using ideas discussed this week.

For example, you might improve the prompt by adding:

  • a clearer task
  • a role
  • more context
  • an explicit output format
  • step-by-step instructions

Include the revised prompt in your post. Then run the new prompt in a generative AI system and include a short excerpt or screenshot of the new response.

Step 3: Compare the results

Briefly explain the differences between the original and revised outputs.

You may consider questions such as:

  • Did the revised prompt produce a clearer or more useful result?
  • What change in the prompt seemed to matter most?
  • Did the improvement come from adding context, specifying the format, or something else?

Explain your reasoning.

Step 4: Remaining limitations

Even improved prompts may not fully solve problems such as incorrect information or vague responses.

Briefly discuss:

  • what limitations or problems still remain in the AI response
  • what information you would verify before using the output in a real task

Expected outcome

Your post should demonstrate:

  • Understanding of how prompt wording influences AI outputs
  • Ability to revise a prompt to improve results
  • Awareness that prompting can improve responses but does not eliminate the need for human verification