Generative AI - Prompt Engineering
Generative AI - Prompt Engineering
Embedded Files
This article highlights a sample of how Generative AI prompt engineering can be used to create scenarios for an instructional design project. It demonstrates how prompts can be refined to add clarity to requests and derive an accurate output in the desired format with relevant details.
Image of Generative AI prompt & response - Revision 1
In the first prompt requested to the AI, I specified the Persona i.e., clearly and directly stated the role that the AI will be taking on: 'You are an instructional designer...'.
I then stated the Objective: '...creating a practice activity in an action mapping project'.
I mentioned the Audience, i.e. who will be the end user of this learning experience: 'In this project, teachers practice...'.
I added Context by mentioning details related to the task such as '...how to encourage a shy student in the classroom to speak up'.
Finally, I set Boundaries by adding limitations to the response, such as, 'Create and list a scenario, provide 3 choices and corresponding consequences'.
This AI prompt has all the basic components of a well defined prompt and generates a fairly accurate response.
After going through the response, I wanted to ensure that the scenario needs to be made more realistic, so I decided to request a conversational format to be included. I stated clearly how long the dialogue should be and how the scenario should end with a direct question to the teacher.
I also requested the Choice and Consequences to also be presented as dialogues.
The response I received was much more realistic and closer to my expectations. The Scenario included a short student-teacher conversation with an ending question, and the Choices and Consequences also contained realistic dialogues.
Image of Generative AI prompt - Revision 2
Image of Generative AI response - Revision 2
One aspect I wanted to improve in the previous response was that after every Choice, the response stated the intent of that choice. For example, the first choice stated 'Directly ask Alex a question during a class discussion'.
I wanted the choices to remain subtle so a learner can actually think through the responses and select the one they feel is most appropriate. So I made a further refinement in my prompt engineering by adding 'For each choice, remove the first sentence explaining the concept'.
I noticed that nothing changed in the response I received.
Image of Generative AI prompt - Revision 3
Image of Generative AI prompt - Revision 4
I refined the prompt further by adding, ''For each choice, remove the first sentence explaining the choice'.
Now I got the exact response I was looking for, with the Choice explanation removed.
This highlights the level of accuracy and refinement needed in prompt engineering to evoke the exact responses we require from the AI.
Page updated
Google Sites
Report abuse