Skip to Main Content
Memorial University

Memorial University Libraries

Generative AI

A guide exploring generative AI tools, how they work, and how to critically evaluate their use.

Considerations for use

Generative AI tools are powerful, and they can be helpful for some tasks. They are not always the most effective or efficient tools for every task, and different tools have different strengths and weaknesses. Furthermore, because of generative AI tools’ capacity for hallucination, a certain degree of expertise and background knowledge is required when using these tools. 

Whenever you are deciding to use a generative AI tool, carefully consider your task, your expertise, and the tool using the strategies below.

Check with your instructor

If you are working on a class assignment, check your syllabus and/or discuss the permitted uses of generative AI tools with your instructor. Different courses may have different rules and requirements for the use of such tools dependent upon learning outcomes.

Remember, university is an opportunity to build your critical thinking skills and subject expertise; these cannot be outsourced to generative AI tools and are prerequisites for using generative AI tools well.

Decision tree for appropriate use

When deciding if it is appropriate to use a generative AI tool, the following decision tree may be helpful as a guide.  

(Faculty may wish to substitute mentions of a “course” with considerations of journal publishing policies.)

A decision tree for using AI in course work. Are you allowed to use genAI? If yes, proceed. Are you using it ethically, or is it taking the place of your own learning? If ethical, proceed. Is it appropriate for genAI, or better to do yourself? If genAI can do it without affecting your learning, proceed. Do you have expertise to confirm the accuracy of the genAI output? If yes, proceed. Will the prompt include personal or copyrighted information? If no, use the tool.

This decision tree was derived from "Should I use AI for this?" created by Jessica Kiebler, Pace University. Modified from the original by Wendy Rodgers, Memorial University. This material is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. 

The ROBOT test: Evaluating generative AI tools

The LibrAIry has developed the ROBOT test, an acronym that you can use to help you evaluate multiple aspects of generative AI tools. ROBOT stands for: 

Reliability of information 

  • How reliable is the information you can find about the generative AI tool? 

  • Is it written by a third party? What are the author’s credentials? Can you detect any bias? 

  • Is it written by the owning company? How much information are they making available? What information is unavailable (e.g. due to trade secrets)? Can you detect any bias? 

Objective of the generative AI tool 

  • What is the goal or objective of the company who created the generative AI tool? 

  • What is the purpose of the generative AI tool? What is it meant to do? 

  • In what context is the generative AI tool designed to work? (e.g. academic research? Public use?) 

Bias of the generative AI tool 

  • What training data was used for the generative AI tool? What bias might exist in this training data?  

  • What biases have been documented in the generative AI tool’s outputs? 

  • By the source of information? 

  • By the party responsible for the generative AI? 

  • By its users? 

  • What biases do you notice in the tool’s outputs? 

Ownership of the generative AI tool 

  • Who is the owner or developer of the generative AI tool? 

  • Who is responsible for it? (e.g. a private company? The government? A think tank or research group?) 

  • Who has access to it? 

Type of AI tool 

  • What type of AI tool is it? (See The LibrAIry’s AI Family Tree for specific types of AI and an explanation of the interplay between AI, data science, and machine learning) 

  • What kind of information system does it rely on?

The ROBOT Test is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. It has been adapted for clarity, inspired by the adaptations made in the University of British Columbia Library’s Generative AI and ChatGPT guide.