When deciding whether to use generative AI for your research as part of your coursework or studies at York University, a final factor, in combination with other considerations we've talked about, is whether or not you can verify what you are finding and using.
In many ways this is about engaging in critical thinking that is a key part of any good research practice and about adopting a wholistic approach to research.
Compare and verify: Consult multiple sources of information to verify or corroborate an AI tool's claims This is necessary as these tools may generate inaccurate information as explained under the Limitations section of this guide. The choice you make about what resources to use to do this are often context-dependent, e.g., the nature of the research assignment and project and the sources an instructor recommends you use. In most cases, even where your instructor allows the use of generative AI, it's unlikely they would want it to be used as the sole information source. Moreover, your assignment will be of a higher calibre, where you have verified the outputs sourced using generative AI. More specifically use these strategies to find other sources to help you test the information you have found using generative AI. This includes ways of finding scholarly information but also other information types:
|
|
Check citations for hallucinations (or fake information) While citation accuracy is evolving and improving with some tools, we advise verifying citations outputted by generative AI tools. See the Limitations section of this guide to learn more about why these issues with citations happen. To verify a citation that is generated by generative AI tool:
|
|
Cite information you source using generative AI tools This is related to the process of verifying what you find. You may be citing what generative AI outputted directly or you may be citing a citation that it tracked for you, that you then had to verify (as per the above point). Either way you need to cite sources accurately according to the citation style you are using. Consult the Citing Generative AI Tools section of this guide for full guidance and instructions on important considerations here. |
|
Consider consulting with us at the Libraries Do this to learn strategies to verify citations you have found using a generative AI tool, or to get help with finding other authoritative sources of information verify or to corroborate what you have found using such a tool. To do this:
|
|
Keep a log of the prompts you use This is a good strategy whenever you want to be able to verify for yourself or anyone who asks, e.g. an instructor, how you approached your research (note too that quite a few tools may keep this history for you, especially where you have an account with them). Some instructors may want you to include a summary of your prompting strategy as part of your assignment in combination with citing generative AI tools, where this is permitted. It can also help you retrace your steps later on, e.g., as research is an iterative process, you may come back to engage in fresh methods of searching or inquiry later, and having a record of your prompts, where using a generative AI tool, will help. Consult the Effective Prompt Strategies section of this guide for more relating to the iterative nature of prompting and generative AI tools. |
|
Where publication of your work may be an end goal, review the requirements or policies that apply for generative AI usage with the publication in question Clearly there will be many frustrations, where you set out to publish and omit to check what applies, and end up finding what you have done, is not a fit, in light of the policies that apply. This is true even for publishing in a student-run journal or a campus newspaper, but will also apply for scholarly journal or book publications. These policies are increasingly available as part of the standard policies that publications issue, and will help you verify if generative AI tools can be used and how their usage should be documented. |