Skip to Main Content

Biology

AI academic search engines

How are AI academic search engines different from other generative AI?

 

Chart showing differences between general LLMs, AI search engines, and AI academic search engines

 

 

 

 

 

 

 

 

 

 

 

CC BY-NC4.0. Erica Nekolaichuk and Kaitlin Fuller, Behind the Buzzwords: What AI Search Engines are Really Doing. Presentation to CAAL-CBPA, August 13, 2025.

 

AI tools for literature searches

A few tools and when to use them

Please always work within the guidelines provided by your instructor for using an AI tool in your assignment. Refer to the library's Generative Artificial Intelligence guide for more information about using AI ethically and critically in your work, including how to use it for research purposes and how to cite it.

The landscape of generative AI is rapidly changing; the following list is only a very small sample of what is currently available. The selections will be monitored and updated as we learn more about them, or better tools emerge.

The tools below were chosen because they:

  • are grounded in scholarly/academic content, unlike general LLM chat, so there is less chance of hallucination/fake citations
  • have free options (with some limitations)
  • search using natural language (questions, rather then keywords and Boolean operators)
  • have relatively user-friendly interfaces

 

Elicit

Highlights:
  • searches Semantic Scholar content
  • helpful documentation
  • keep selected and uploaded articles organized in a library
  • pulls from its own source material, so lower risk of hallucination
  • search options available with free account
    • Find Papers retrieves up to 50 articles based on research question and summarizes the most relevant, customize results with filters and compare across methodology, outcomes, etc.
    • Chat with Papers allows you to select papers or upload additional ones to your library and Elicit will provide you with prompts (subject to plan)*
    • Upload and Extract other article pdfs to summarize (subject to plan)*
Limitations:
  • results appear in table format and are not accessible with a screen reader
  • uses natural language searching so results may not be reproducible

Best use: developing a research question, finding gaps in the research

 

Remember to operate within copyright guidelines. To date, there is still some ambiguity about what you may or may not upload to one of these tools, for example, when using a feature such as Elicit's "Chat with Papers". Our suggestion is that users ONLY upload open access articles which have a Creative Commons license clearly displayed. 


The-Literature.com

Highlights:
  • searches only on PubMed, so very relevant for Health and Life Science topics
  • works best when question is framed in PICO
  • answers in a variety of formats including plain language synopsis and critical review, sometimes including a "TLDR" summary
  • some searches offer an "iterative query" for further shaping your question
  • developed by a physician for rapid evidence-based answers to clinical questions
  • free, does not require login
Limitations:
  • not reproducible, meaning you won't get the same results for a given query, because it is using semantic searching rather than matching keywords
  • no supporting documentation or Help guide

Best use: for research topics in health and life sciences, quick answer to a clinical question when a comprehensive search is not required

 



Semantic Scholar

Highlights:
  • Searches its own corpus of approx. 220 million articles
  • Uses AI to extract meaning from articles, "Ask this Paper" feature, "Explore topic" from some pages
  • Identifies influential and related articles, filtered citations allows researchers to view citations specifically related to methods, results, etc.
  • Some articles have a "TLDR" summary
Limitations:
  • Smaller data training set than other tools, so may not be comprehensive
  • Majority of collection is science focussed

Best use: Primarily useful for research and discovery; citation tracking

Caution!

The tools above are NOT a replacement for doing your own thorough and systematic searching of the scholarly literature. As yet, there is no generative AI tool that will confidently and consistently find you all the relevant literature for your search. In fact, according to a recent blog post, a few major academic publishers have recently pulled their abstracts and full text articles from the foundational data that some of these tools are trained on, which in turn limits the degree to which your searches will be comprehensive if you only employ them in your search.

At best. these tools and others like them may be considered as aids or supplements to your research process, and may help you with framing your question or leading you to some articles you might not otherwise find.

 

Questions?

If you are aren't sure about the quality or credibility of the articles you discover by using these tools, here are a few things you can do:

 

With thanks to Ontario Council of University Libraries' AI Tools for Academic Libraries, August 18, 2025.. This blog series provides many more suggestions for AI tools and research assistants appropriate for students and researchers in academic libraries.