Skip to Main Content
My Library Account

Generative Artificial Intelligence

Tips for Prompting (click to expand)

Context

Providing relevant context can help AI understand the task better, such as explaining the roles of both AI and the User in the task, outlining your requirements, and specifying the anticipated results. With a given role the AI model knows whether it should act as a knowledgeable assistant or any other specific character.


User:

I am a college student preparing for a debate on artificial intelligence's impacts on employment. Please provide a step-by-step guide to help me get started.

AI:

Certainly. ...


Or try this one:


User:

I am preparing a college debate on AI's impacts on job market. My argument is that AI will lead to job loss. Please provide an outline down to second-level headings.

AI:

Certainly. ...

Instructions

Include explicit instructions for the desired behaviour, for instance, asking the model to think step by step, rephrasing, searching, summarizing, synthesizing and so on.


User:

Please find some academic papers on the various ways that AI is affecting the job market. Compile a list of primary opposing viewpoints: AI leads to job reduction, and AI has the potential to generate new employment opportunities.

AI:

Of course! ...

Examples

Providing example inputs can help the model understand the desired patterns or formats for its responses. Example inputs may include the use of examples, demonstrations of the expected behavior, a set of relevant keywords or descriptors, the abstract of an interested article, a reference text and so on.


User:

I found a paper that talks about Climate Change impacts on Canadian cities. Please list a few more articles that address how Canadian cities might be impacted by Climate Change.

AI:

Sure. ...

Outcome Indicators

Direct the model to generate a response for a given style, tone, format or scenario.


User:

Assist me to finalize the gathered data. Divide them into four parts: Opening Statement, Supporting Arguments, Possible Counterarguments, and Conclusion restating our main points and emphasizing why our team's stance is more convincing.

AI:

Absolutely! ...

Testing & Tuning

The process of prompt creation often involves lots of testing and refining of prompts, and the effectiveness of prompt techniques may differ depending on the particular task, language model, plugins / extensions/ sources employed and the user's prompt engineering experience.

Tips & Tricks

  • When crafting a prompt, use plain English and be specific about what you want the tool to do.
  • Providing background information, context, or examples can be helpful.
  • Divide a complex concept into smaller prompts.
  • Ask ChatGPT to act as an expert to explain a complex concept step by step.

    • Common Types of Prompts

    Action Prompts

  • Can you write / explain / summarize / provide / analyze / compose / re-write /translate ...
  • Give me a table summarizing some of the key differences between ... and ...
  • Rephrase this sentence: ..... / Expand the idea based on the following descriptors .....

    • WH Prompts - (what, when, where, who, which, whose, why, how...)

  • What are the leading causes /factors / impacts / definitions / benefits of challenges, barriers, issues with .....
  • What would be a good image to use for my talk about plagiarism?
  • How to search / cite / code / convert (e.g., the data into a table) ...

    • Prompts To Set A Specific Context

  • Summarize the plot of ... in one short paragraph.
  • Create a list of five ChatGPT features that can be used to promote...
  • I will have a debate about the use of standardized testing in schools. Can you help me find some arguments in favor of this practice, as well as some counterarguments?

    •  

    Adapted from "AI-Based Literature Review Tools" by Texas A&M University Libraries: https://tamu.libguides.com/c.php?g=1289555&p=9642751