Generative AI and Research

When you ask a generative AI tool for information, can that information be used as part of your research? What about sources that may incorporate AI-generated information but are human-authored? 

These are difficult questions to answer but it may help to think about the most common roles that sources play in our our research: filling a gap in our knowledge, confirming something we already know, engaging with a conflicting perspective, and contributing to a scholarly conversation. 

Let’s think about how AI may or may not fit into these different roles. 
 

When you’re doing research for academic work, you are often working with a topic you are still learning about. But even if you’re using a topic that you’re quite familiar with, there might still be gaps in what you know about your topic. Before you write about or present on a topic, you want to make sure that your knowledge is well-rounded. That means you want to fill in at least some of those gaps. 

If you ask a chatbot like ChatGPT to tell you more about a given topic or give you an overview of a topic, the response it generates may certainly include information you didn’t know before. If so, that’s great!

So: can you cite this information as a source? 

Here are some things to keep in mind: 

• You generally can’t cite an AI tool as a source of information the way you would a book or an article. 
• There may be some exceptions to this, but they are all based on your professor’s preferences. So before citing an AI tool as a source, you’ll want to consult with your professor.
• Instead, the best thing to do is use the information the AI gave you as a starting point. You’ll want to find outside sources like books and articles that support the information the AI has generated for you. These are sources you can cite. 
• If you need help finding sources that support what an AI has generated for you, the AI can help. While it won’t tell you the exact sources it’s basing its response on, it will give you a general sense of where and how to search for sources that support its response or possibly recommend specific sources that you can then search for online or in a library’s database. 
• As always, be cautious. By now, we all now about AI’s tendency to hallucinate. The information it gives you may not be accurate and if you ask it for specific sources, those sources may not really exist. That is why searching for the actual sources is important—it verifies the information and the existence of a particular source. 
   

If you are researching a topic with which you already have some familiarity, chances are there are things about it that you already know. So why do you need to cite sources that show this? 

Citing sources to support something you already know is a way to lend credibility to an argument or other piece of research or writing. Being able to point to a source that supports your existing knowledge shows that what you are stating or arguing is not simply a personal opinion or that you made it up. 

If you ask an AI tool to summarize a topic for you and the summary it responds with confirms what you already know, can you cite that as a source? 

Again, here are some tips: 

• Citing an AI tool as a source is not a good way to lend credibility to your argument or perspective. That’s because AI tools themselves often hallucinate or make things up. We also have no way of knowing what sources the AI tool is basing its response on, which makes it difficult to confirm the credibility of the information it generates. 
• Citing outside sources, including books and articles by known experts or authorities, is a better, more accepted way to be viewed as credible because those books and articles have often gone through a publishing and/or peer review process and the authors who wrote those sources had to cite where they got their information, which gives them credibility. 
• Again, AI can help you by recommending where or how to search for sources to confirm a particular piece of information or even recommending specific sources. But AI’s tendency to hallucinate means that you will still need to verify its recommendations, including whether or not an expert it names on a particular subject is a real person and whether a source it cites is a real source. 

In academic and scholarly work, it’s important to engage with rather than avoid conflicting perspectives. While it may seem that the existence of a conflicting perspective damages your own argument, engaging with that perspective in a thoughtful way instead of ignoring it or dismissing it actually lends credibility to your own work and makes your research more well-rounded. 

Here are some tips: 

• You can use a chatbot or similar tool to practice engaging with a conflicting perspective by prompting the tool to adopt the persona of someone who disagrees with you and then engaging it in a conversation. If done well, this can help you understand where the points of disagreement might be on a topic and think about how to address them. 
• You can also tell an AI tool about your research and ask it to identify potential conflicting perspectives that you may need to be aware of when making your argument. 
• Can you then cite that conflicting perspective as a source? No. As usual, it’s better to use AI as a starting point that can help you become aware of what to look for and then search for outside sources that represent the perspectives you’ve learned about. Citing these outside sources is more credible than citing an AI tool. 
   

The goal of scholarly research is to build on an ongoing scholarly conversation about a given topic. That may sound confusing if you’ve never done it, but as you become more advanced in your academic research, your own research will require you to become aware of the scholarly conversation in a particular field and all of the perspectives represented in it and think of ways to then contribute something to that conversation. 

• AI can help you become more familiar with the scholarly conversation in a field by providing an overview of that conversation. 
• An AI tool can also tell you who the most well-known experts are in a scholarly conversation and what they’ve researched. 
• Some AI tools can help give you a sense of the scholarly conversation by showing you the connections between one scholarly article and other articles in the same field of research. 
• AI tools can also help you brainstorm ways to contribute to the scholarly conversation yourself by suggesting research questions and potential gaps in a field of knowledge that may need further investigation. 
• Keep in mind that though AI can certainly be useful for these activities, some of these areas (particularly the first two bullet points above) are ones where AI tools are especially prone to hallucination—making up information or names of people who don’t exist or misrepresenting someone’s research. As always, any information that AI gives you is a good starting point but you will want to independently verify it yourself using outside sources.