ENG 123 7.F: A Quick Guide to Prompt Engineering

Today’s Plan:

• Proposals
• Resource
• A Quick Guide to Prompt Engineering
• For Next Week

Proposal / Resource

From McIntyre and Fernandes,First, a list of Resources on Refusing, Rejecting, and Rethinking Generative AI in Writing Studies and Higher Education

A Quick Guide to Prompt Engineering

As you are aware, I’ve been doing quite a bit of research and writing with ChatGPT this semester. And, as you are also aware, I’m not necessarily incredibly happy about doing that research. But… Many of your research proposals are incorporating ChatGPT into your project. To help facilitate that work, I would like to go over some fundamental principles for working with ChatGPT. I have about 20 minutes to write this, so let’s see what I can do in twenty minutes. First and foremost, remember Ethan Mollick’s advice: “treat AI like a human but tell it what kind of person it is” (Co-Intelligence. That is, do not sit down and merely ask it to do something. Sit down and have a conversation about what it is you want to do. Let’s examine two sample prompts Mollick shares in the book:

• You are an expert at marketing. When asked to generate slogan ideas you come up with ideas that are different from each other, clever, and interesting. You use clever wordplay. You try not to repeat themes or ideas. Come up with 20 ideas for marketing slogans for a new mail-order cheese shop, make them different from each other, and make them clever and creative.
• I am stuck on a paragraph in a section of a book about how AI can help get you unstuck. Can you help me rewrite the paragraph and finish it by giving me 10 options for the entire paragraph in various professional styles? Make the styles and approaches different from each other, making them extremely well written.

Note: some of what Mollick writes here seems useless and vague to me, but the general principle is smart. Let’s examine how I and some of my students prompted ChatGPT while writing our own papers. What I hope you see from these examples is that getting quality material from ChatGPT requires you teach it how to write. Writing with GPT requires quite a bit of revision (as reprompting), and guiding that revision requires that you already know quite a bit about writing. This is one reason why having a rubric is helpful (whether one I supplied or one you researched and designed)–because you have to know what to tell the machine. Telling it to write something “extremely well” is really, really useless. I think I have found a resource to help with some prompt engineering. I stumbled upon this yesterday while researching how AI is being integrated into college-level writing classes. From Ranade et al, 2024, “Using rhetorical strategies to design prompts: a human-in-the-loop approach to make AI useful”:

I want to look inside the article at their use of Lloyd Bitzer’s (1968) research on “the rhetorical situation,” that will help us parse out the image above.

For Next Week

In your proposal, you developed a calendar of work that has to get done. Start doing work. Next Friday, I will ask you to turn in a link to your week’s work product and give me a quick overview of what you accomplished.