Coming up with a topic and a focus for your research is one of the hardest tasks in the whole process. Of all the possible topics in the universe, you have to find one and you have to find a way to write a solid argument or coherent exploration within a set page limit. If you are totally stumped on topic ideas, AI might be able to help.
Remember:
Research questions can help you focus your topic ideas down to a specific question to address in your papers. For example, if you want to write something about the challenges faced by transgender high school athletes, a research question can focus that massive topic to a more manageable scope: "What are the most effective ways that parents can advocate for their transgender children who want to participate in sports?" Instead of researching and discussing all of the potential challenges faced by this group, we can do a deeper dive into actions parents can take for their children.
We have some tips for getting topic and research help from AI tools like ChatGPT, Bard, and Claude.
Instead of telling the AI to give you topics, tell it to give you a specific number of topics. I asked Bard, "Give me three research questions about how parents can advocate for their transgender children who want to participate in sports" and it suggested:
My prompt included a set number (three) of things I wanted the tool to give me (research questions) and lots of detail about the kind of question I was looking for.
Remember that AI tools don't "think" the way we do. They process incredible amounts of written information to respond to our prompts. While AI can appear to anticipate our needs, the responses they give are only as good as our prompts. The more you refine your prompts, the better the responses will be. You may need to have a "conversation" with the AI to get what you need.
I initially told Bard that I needed to write a seven-page paper about transgender high school athletes and that I wanted it to give me five topic ideas. Bard gave me some very broad topic ideas:
These topics are too big to cover in just seven pages, so I asked Bard to combine two ideas that I thought were interesting: policy and athletic ability. Bard gave me a great topic and potential research question:
Once you have a topic and research question you like, test it out in Fulton Library's OneSearch database or with the search engine of your choice. Before you commit to a topic, make sure there is enough information available to support your argument or topic exploration.
Partially peer reviewed. Some full-text content. Videos. Audio Recordings.
OneSearch is the default search on the Fulton Library’s homepage. Includes the library’s books and videos, articles from many library databases, and more.
Another tricky part of the research process is figuring out what to put in the search box of a library database or search engine. Like AI tools, databases and search engines don't really "think" either. They just look for a match to the letters you type in, in the order that you type them in. Choosing the right words is very important, but if you're new to your topic it can be hard to even know what the right words are. AI tools can help with this challenge.
For best results:
Use the tips and strategies in our Starting Your Research guide to format the words suggested by the AI into searches that will work in the Fulton Library's databases.
Once you have your search terms, you can also use AI to format those searches so that the library's databases can understand them. Bard is the most useful tool for this (as of January 2024).
To get started, use a prompt that instructions the AI to create a Boolean search based on your research question, like this:
Bard will give you multiple options to choose from and we recommend using different searches to make sure that you find as much relevant material as possible. ChatGPT, Copilot, and others might only give one search.
AI tools can be very useful for developing topic ideas and identifying search terms but they should not be used to locate books or articles for your research. This is because AI tools can "hallucinate" information that doesn't exist. When users ask AI tools to find articles, websites, books, or other sources, there is a chance AI can create citations that look like valid sources. However, when you go looking for a copy of that source, you might find the supposed author is a real person, but they never wrote an article with that title or that the article title is a mishmash of the titles of other articles.
If you need help finding sources for your research—or if you suspect a citation is an AI hallucination—ask a librarian! We're always happy to help you find sources and answer questions about research.
AI tools can help you find articles by using the citations at the end of scholarly (peer-reviewed) articles to find additional articles on the same topic. To use these tools, you must start by giving the AI a "seed" article. The AIs listed below will then find as many articles cited in your seed article as it can, plus articles that cite the seed article since it was published. There are some limitations due to paywalls, the ability of the AI to identify citations in PDFs, and the lack of content in some subject areas.
To explain this a little better, I used Semantic Scholar with the following article as my "seed":
Buzuvis, E. (2021). Law, policy, and the participation of transgender athletes in the United States. Sport Management Review, 24(3), 439–451.
https://doi.org/10.1080/14413523.2021.1880757
Here's what Semantic Scholar found for me:
The 21 citations listed all cite my seed article. The 43 references are all cited in my seed article. If I click on the citations or the references, I'll see additional articles on my topic that I might be able to use.
Semantic Scholar and the other AI-assisted databases below are not perfect. They will miss things. For example, the article I used as my seed actually has 48 sources listed in its bibliography and Google Scholar found more sources that cited my seed article than Semantic Scholar did.