In February 2024, I wrote an AJN Off the Charts blog post titled “Leveraging AI and Technology for Comprehensive Research: Tips for Researchers and Students.” Since then, the field of AI has undergone rapid evolution. It is evident to all of us watching the field develop that companies hosting and developing large language models (LLMs) would eventually target scientific research. In my previous post, I explained that there is no single software solution for conducting research or literature reviews using AI. However, the deployment of new features in AI software platforms, such as deep research capabilities, may mislead us into thinking otherwise. The purpose of this blog post is to introduce the idea of deep research tools, while also providing tips for users who wish to explore these evolving tools.

What is deep research?

Image: Marcus Winkler/Unsplash

Deep research is a term used by LLM software platforms that allow users to enter a prompt to initiate an in-depth process that involves finding, analyzing, and synthesizing “hundreds of online sources to create a comprehensive report at the level of a research analysis” (OpenAI, 2025). There is also a consideration for time using this tool, as the responses are not instantaneous and result time can vary based on the topic and the complexity of the research question. While such AI output may sound revolutionary, there are some limitations to share, especially in terms of publishing standards. While the research report produced through deep research tools is typically excellent, I have yet to see a report that meets the editorial standards for publication at a peer-reviewed journal.

One primary limitation is that LLMs remain unreliable without human oversight to verify accuracy and ensure that references meet the quality standards of peer-reviewed academic publishing. The other limitation is that (as of right now), deep research tools search only one scholarly database–Semantic Scholar, or in other cases PubMed. While both are excellent academic databases, searching just one or two databases severely limits the search because it excludes articles that might be indexed on other scholarly databases. Aside from other limitations that will be discussed below, a research report entirely created using deep research tools that rely on one database would not be a publish-ready document. It should be noted that as these tools develop, we will likely see deep research using various academic databases, potentially solving this challenge. So how should we use this new tool in academic research spaces?

Incorporating deep research tools in your research workflow

Is there a place for deep research tools? The answer is yes, but with caveats. In my previous February post, I demonstrated how to make the literature review process easier by piecing together various tools to avoid hallucinations and using AI to refine and clarify—more of a support system or research assistant and less as an ultimate “do it all tool.” I would say the same thing about utilizing the deep research features that are increasingly being integrated into AI software platforms. The tool does a good job of giving researchers or students a high-level view of the research field of their interest.

As an example of a use case where deep research tools may be beneficial, let’s consider a topic such as genetics and genomics in nursing education. A student or novice researcher may be interested in this area. The first step is to determine what has already been done on this topic. This is typically where the literature review as a research tool comes in handy. However, a literature review often requires students and/or researchers to have a well-developed interest in the topic. I view the deep research LLM tools as a way to initiate the process—a pre-literature review, if you will. For example, a user might use the following prompt:

“I am interested in the field of genetics and genomics in nursing education. I would like to get an overview of what has already been published on this topic, while exposing gaps so that I can build my research in this area.”

Obtaining a full report on this topic based on the prompt could serve as a great precursor to undertaking a more formal literature review. From there, the student and/or researcher could utilize the techniques outlined in my February 2024 blog post to leverage AI as a support system for conducting a rigorous literature review. Deep research tools can save researchers valuable time in finding topics that not only interest them, but also where they can build a valid area of research around.

Using deep research tools carefully

As with all AI/LLM products on the market, researchers should exercise caution when interpreting the results obtained from deep research reports. It is crucial to verify the accuracy of all sources in the report before proceeding with more formal research processes. The report cannot serve as a standalone literature review and is not suitable for publication. Recognizing the value and pitfalls in these tools requires a high level of clinical reasoning and human oversight.

Used correctly, these evolving tools can save time and help users refine potential topic ideas while eliminating others. Many students and researchers alike undertake extensive literature reviews only to find that what they are interested in has already been done. The time it takes to pivot and start the work over again can be exhausting and lead to incomplete work. Using deep research tools with appropriate caution has the potential to mitigate these common academic challenges for certain research purposes. Other limitations remain even with the resolution of these glaring limitations. For example, no amount of database searches can take the place of a human researcher.

Bottom line: proceed with caution and keep an eye on the trends in this area.

By Justin Fontenot DNP, RN, NEA-BC, FAADN, associate professor, Tulane University School of Medicine, Program of Nursing.