Top tips for double checking AI info in your academic research

As nearly a quarter of adults in the United States regularly use artificial intelligence chatbots, a 2026 Pew Research Centre study found, it is vital for students to ensure that the information they receive from these tools is accurate for their academic research. Not all AI tools are created equal, so selecting the appropriate one for a specific task is essential. Two Coastal Carolina University librarians provide four steps to help students double-check AI-generated information.

First, choose the right AI tool for the task at hand. Different tools, such as ExLibris’ Primo Research Assistant, Research Solutions’ Scite, Anthropic’s Claude, and OpenAI’s ChatGPT, have different datasets and algorithms, so they may not be interchangeable. Some tools may be better at answering certain types of questions or producing different kinds of responses, so it's essential to consider which tool will best suit the needs of the research process.

For instance, ChatGPT can provide overviews and summaries at the beginning of a research project, while Primo Research Assistant or Scopus AI can help find and examine peer-reviewed research.

Second, ask follow-up questions to evaluate the AI-generated answers. Even if the initial answer seems correct, it's essential to dig deeper to understand where the information came from, the AI's confidence in its accuracy, and any missing perspectives. It's also crucial to be aware of potential biases and ask the AI to explain its reasoning for the generated answer. Additionally, make sure the prompt is neutral, avoiding leading or emotionally charged language.

Third, trace any claims made by the AI tool to its original sources. Even if the AI provides a citation, verify whether the information is used in the correct context and whether it comes from a credible source. If the AI fails to cite sources or if the cited sources are unreliable, conduct further research using library databases or other trustworthy sources. In some cases, it may be more efficient to rely on library resources instead of AI tools.

Fourth, be aware of less visible issues when using AI tools, such as missing perspectives or biases in the generated information. AI tools primarily focus on freely available and English-language sources, so it's vital to consider whether expert knowledge might be absent from these datasets. Supplement AI-generated information with expert viewpoints when necessary. Lastly, always cite the AI tool used if quoting or paraphrasing generated text in your academic work, just as you would cite any other source.