Artificial Intelligence

The Oxford English Dictionary uses the following definition for the term "intellectual property": Property (such as patents, trademarks, and copyright material) which is the product of invention or creativity, and does not exist in a tangible, physical form.

When someone uses AI, it brings to mind the following question that many have been asking is: Who is the copyright owner of the works that are AI generated? In Canada, this is a question that the Federal court is being asked to determine, with regards to whether AI can be the author of a work. As more cases continue to make their way inside the courts in Canada, it is inevitable that we will eventually see clearer language regarding the authorship of works which utilize AI and how these tools will affect copyright. 

The United States Copyright Office has published a guide for works containing material generated from Artificial Intelligence. In short, whether an item may be protected by copyright can be a bit murky, and depends on the amount of contribution from the human authors and from the AI tools that were utilized. One example included to illustrate this is that a graphic novel with human-authored text would have its textual components protected by copyright, but the AI-generated images themselves (the non-human component) would not be protected by copyright. While these guidelines are not necessarily applicable to Canadians, it is likely that some of these cases will be referred to, and shape the future of AI usage and intellectual property for us as well. 

What this means for you:

• Inclusion of AI in your work may potentially prevent you from protecting your work through copyright.
• You should always make it clear how you have used AI tools in your work, and what your contributions are (always give credit and cite your work!).
• Because generative AI models save user data to train Large Language Models (LLMs), you are generally not allowed to upload copyright protected works (such as library licensed resources) into these AI models. 
  • One exception is that you can use Microsoft Copilot, which is Trent University's only approved generative AI tool for use. Under our use agreements with Microsoft, Copilot will not keep, train, or view any data that you upload.
• You should familiarize yourself with publishing and publisher policies if you plan to use AI tools in any work that you are considering publishing (such as manuscripts you are planning to submit to an academic publishing journal). Below are examples of publishers that have language regarding the acceptability of AI tools, and how to credit AI tools if you used them:
  • Cambridge University Press Authorship and Contributorship Guidelines
  • Elsevier Policies - The use of generative AI and AI-assisted technologies in writing for Elsevier
  • Science Journals Editorial Policies
  • Springer Journals Artificial Intelligence (AI) Editorial Policies
  • Taylor & Francis AI Policy
  • Wiley Generative Artificial Intelligence Guidelines for Authors

If you require assistance, or have questions pertaining to the usage of AI in your manuscript for a work that you intend to publish, please reach out to your librarian for further assistance. You may connect with us at library@trentu.ca

When it comes to using AI, it is important to remember that varying tools will work best for completing different tasks. For instance, while spoons are great tools for eating certain types of food, you would much rather use a fork when eating spaghetti. The same logic should be applied when determining which AI tool you plan to employ during your research or writing processes.

Here are some tips for you to make the most out of using AI tools to meet your research needs:

• Brainstorming: Text generators like ChatGPT excel at helping you to brainstorm ideas for research questions because it can offer a wide range of possibilities that you may not have thought of. It is also great for identifying starting points for finding more resources, since it can quickly process and suggest a broad spectrum of information. 
• Finding search terms and synonyms: In particular, Generative AI tools can help you to find search terms. Search terms are words related to your research topic that you can use to search for relevant sources in Omni, Trent's library search tool. You can broaden your search for research sources by using AI to create lists of synonyms and related search terms that you can search in Omni to expand your research scope or explore your topic in greater depth.
• Know the limitations of the tool you are using: Know what some of the main strengths and limits of the tool you are using are. There are tons of resources that you can explore to learn more about AI tools. Interestingly, with some text-based AI models (like ChatGPT), you can simply ask them what their limitations are and it will tell you some of its weaknesses.
• Write great prompts and modify as needed: The output of AI tools are only as good as the information that they are being fed. There are many ways in which you can optimize your prompts, to deliver better results. For instance, providing more context is better than less. Continuously modifying the outputs by keeping the conversation going is another good way to tailor your results. Providing examples before making a query will ensure that the outputs are closer to what you had envisioned. There are many more examples of ways that you can improve your prompts. Here are some great resources that you can read, to learn more about optimizing your prompts:
  • Learn Prompting: Your Guide to Communicating with AI
  • Getting started with prompts for text-based Generative AI tools
  • 6 ways to write better ChatGPT prompts - and get the results you want faster
  • Using ChatGPT effectively

AI Tools - Links to Standalone Platforms

As mentioned in the introductory portion of this research guide, there is an enormous array of AI tools that can be used for diverse purposes. Some of these tools are highlighted below. Please note that some of these tools cost money, while others are free to use, and some require you to sign up.

Examples of text generators include: 

• Copilot (this is the only Trent approved generative AI tool for use)
• ChatGPT 
• Gemini
• Claude

Examples of image generators include: 

• Firefly
• Midjourney 
• DALL-E 3 
• Stable Diffusion

Examples of audio generators include:

• Otter
• Parrot

Examples of research-specific generators that can help you with scholarly sources of information include:

• Elicit
• Scite Assistant
• Consensus
• ResearchRabbit

AI Tools - Built Within Research Database Platforms

Many AI tools are now being directly integrated within research database platforms. As a Trent community member, you get access to a multitude of databases for performing research. Many of these databases, such as ProQuest One Business has a research assistant that can provide you with key takeaways, suggest other relevant resources, and potentially improve your searches by suggesting more relevant keywords. JSTOR is another database with an AI tool (currently still in beta mode) that is intended to help improve users with meeting their research needs.

When it comes to using these tools, most are built in ways such that they don't have the same shortcomings as other popular generative AI tools, which may display bias, inaccuracy issues, etc. As such, you are recommended to take advantage of these tools to improve your research, as these tools will often make the process of gathering relevant documents a more efficient process. 

If your instructors approves the use of AI tools when conducting your academic work, you should always double-check your information and cite any informational sources that are not your own, and information that is not considered common knowledge. You should never pass on work as your own, when you are using someone else's words, even if those words are from AI generators. Unauthorized use of generative AI to write your course assignments for you puts you at risk of committing plagiarism, which could get you into academic trouble. Or, as mentioned in the "Intellectual Property" section of this research guide, using generative AI tools could even disqualify you from submitting a manuscript that you intend to publish. 

Please take a moment to familiarize yourself with the academic integrity policy for undergraduate students and the academic integrity policy for graduate students. Both of these policies contain language pertaining to the usage of AI generators in your coursework. 

AI models are trained on a variety of sources, but many of these websites where information is obtained from are of varying quality and can emphasize specific viewpoints, resulting in bias in the AI's responses to inquiries. Without proper quality controls in place, there have been many known instances of generative AI tools perpetuating stereotypes, along with being racist, sexist, and homophobic. For instance, Stable Diffusion believes that the world is predominantly run by white males, women with dark skin tend to flip the most burgers and women rarely obtain high-ranking positions such as doctors, lawyers, or judges. Even when there are attempts made to improve AI models, there are problems that result with addressing the racial bias problems that are present in many AI generators. For instance, Google's Gemini portrayed Nazis as being racially diverse, when trying to be more diverse in its image outputs. 

There is much work that needs to be done in order to improve the biases in outputs of generative AI tools, and it important to be conscious of these shortcomings when you use these tools during your research endeavors.

There are many tests that have been developed to evaluate AI tools. The ROBOT Test, developed at McGill University prompts you to assess the following dimensions of AI tools: Reliability, Objectivity, Bias, Ownership and Type.

Reliability

• How reliable is the information available about the AI technology?
• If it’s not produced by the party responsible for the AI, what are the author’s credentials? Bias?
• If it is produced by the party responsible for the AI, how much information are they making available? 
  • Is information only partially available due to trade secrets?
  • How biased is they information that they produce?

Objective

• What is the goal or objective of the use of AI?
• What is the goal of sharing information about it?
  • To inform?
  • To convince?
  • To find financial support?

Bias

• What could create bias in the AI technology?
• Are there ethical issues associated with this?
• Are bias or ethical issues acknowledged?
  • By the source of information?
  • By the party responsible for the AI?
  • By its users?

Owner

• Who is the owner or developer of the AI technology?
• Who is responsible for it?
  • Is it a private company?
  • The government?
  • A think tank or research group?
• Who has access to it?
• Who can use it?

Type

• Which subtype of AI is it?
• Is the technology theoretical or applied?
• What kind of information system does it rely on?
• Does it rely on human intervention? 

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

To cite in APA: Hervieux, S. & Wheatley, A. (2020). The ROBOT test [Evaluation tool]. The LibrAIry. https://thelibrairy.wordpress.com/2020/03/11/the-robot-test

There are also other tests that you can use to evaluate generative AI. For example, the CRAAP Test, which stands for Currency, Relevance, Authority, Accuracy and Purpose can help you determine if using a specific AI tool is appropriate for meeting your research needs.

When using most generative AI tools, your chat data is saved, and used to improve the AI model. Thus, it is important not to share any sensitive or personal information when using these platforms. Unfortunately, even if you decide not to share confidential information with AI, your information may still be getting sold or harvested by AI. For example, the New York Times claims that ChatGPT scraped data from millions of its articles and infringed its copyright. Additionally, Taylor & Francis recently sold access to their research to Microsoft AI (without the authors of these papers knowing, or being compensated for their contributions).

It is also worth repeating that Microsoft Copilot, is Trent University's only approved generative AI tool for use. Under our use agreements with Microsoft, Copilot will not keep, train, or view any data that you upload. Utilizing Copilot will ensure that your privacy is maintained, and that your data is secured from theft, corruption, and unauthorized access from parties with nefarious intentions.