A Lecturer’s Guide to AI Prompting

As a Senior Lecturer actively using AI tools, I am interested in seeing the concrete impact that AI has on my work. In this article, I will delve into my experiences with prompting, a crucial skill in the evolving landscape of AI development.

In my work, AI is used for different purposes, e.g., brainstorming on different ideas, creating drafts, auditing course content and criteria, creating diverse types of assignments, and editing text. I mainly use Microsoft Copilot and ChatGPT for these. Mastering prompting skills requires continuous development, and I do this by experimenting with different scenarios and by exploring materials, providing help to learn the basics and better understand the logic of a prompt.

Depending on the purpose, the prompt can vary in length from concise to detailed. The quality of the outcome received depends on how well-crafted the prompts are. When providing minimum effort prompts, the outcome received will most likely be of a low quality. Planning the prompt is an important phase and as in all other cases, the time invested in planning will save time on finalizing it. Random are the cases in which the outcomes offered by the chosen AI tool are ready to use in the first round. Usually, the interaction with the AI tool continues until the outcome closely aligns with the expected outcome. In addition, the outcome always requires human editing.

Prompting in Action: A Practical Example

Let us explore an example of how a lecturer can use AI to create a lesson plan for a course topic. Despite the increasing use of AI tools, most of these still do not know the user. Therefore, the first step in prompting is to introduce yourself or assign your role to the AI. Here are some key points to include for the AI (adapted from Wharton School, 2023):

• The role that you are expecting it to perform. By doing this, you provide a context and a perspective to take into consideration.
  E.g., “You are helping a lecturer with a lesson on creating a user-friendly interface.”
• Let AI know if students are familiar with the topic or if it is new to them. If students are already familiar with the topic, provide a summary. This will help the AI determine the starting point and decide whether it can skip the basics.
  E.g., “The students are first-year university students and do not have any kind of prior knowledge and experience on the topic. The students are currently studying Web Development course and creating user-friendly interfaces is one of the main topics of this course.”
• Add the aim for the students to learn or achieve by the end of the lesson.
  E.g., “By the end of the lesson, the students will understand the main facts in how to create a user-friendly interface.”
• If needed, add constraints to control the text style for your prompt.
  E.g., “Use formal academic style.”
• Inform AI if you are aiming for a lesson plan that incorporates diverse teaching methods or if you are interested in using a specific method, e.g., some kind of engaging activity.
  E.g., “During this lesson, the students are expected to work in pairs and utilize the peer review method.”
• If a specific format for submitting lesson outcomes is required, make sure to inform the AI accordingly.
  E.g., “The students are required to utilize the Pair Review activity available on the course Moodle platform. Each student must evaluate their fellow student’s work, assign points, and provide a justification for their assessment. Additionally, students are expected to conduct a self-review of their own outcomes.”

This example illustrates just one way to create a prompt on a specific topic. There are many different methods for prompting, each potentially producing different responses. Additionally, the choice of AI tool can influence the outcomes of a prompt. When struggling with prompting, one effective approach could be to consult AI itself for guidance on how to prompt on a specific topic. When prompting, we are interacting with the AI, and in my experience, the interacting process is more important than a single prompting skill.

Providing examples helps AI to better understand the goal, and by providing clear steps, it helps AI split the topic. As generative models advance, the techniques for prompting them are continually evolving. It is evident that educational practices will evolve and prompting techniques will change. It is left to be seen what impact these will have on future learning. (Wharton School, 2023)

While completing this article, OpenAI introduced Canvas in early beta. Canvas is a new interface for working with ChatGPT, aiming to enhance the interaction with ChatGPT. We have seen that until now, most text-based AI tools and developing interfaces have been more interactive and intuitive, and they will have an impact in producing the prompts. Most likely, prompts will be more concise, and refining the prompts will happen through functionalities implemented as part of the interface. As OpenAI states, enhancing the user interfaces for an AI tool requires rethinking how we interact with it. (OpenAI, 2024)

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work, the author used Microsoft Copilot to enhance the text’s language and grammar.