Artificial Intelligence (AI)

Using AI For Literature Searching

We would caution you against using generative AI for literature searching as it lacks the authority and integrity of things like LibrarySearch and the specialised databases we provide access to. If you do use generative AI for any aspect of the research process, we'd strongly recommend reading through the page on Using AI Critically.

On this page, we'll be taking a look at search engines and highlight some issues around using generative AI when conducting a literature search.

You must follow the latest guidance on using Generative AI as provided by your School.

Search Engines

It's important to note, that the technology we have used and continue to use before the advent of ChatGPT was not without its problems. Conducting effective literature searching online has always been an important skill to develop, and it requires a good understanding of how search engines function. An effective, high quality literature search requires an understanding of the various operators and how to use them in your search. Some aspects of the process, however, remain a mystery. For example, does the user really know how relevance ranking actually works? Why one article is rated as more "relevant" than another?

In her book, Algorithms of Oppression, Safiya Noble explores how search engines reinforce racism in their outputs. She argues:

"Search does not merely present pages but structures knowledge, and the results retrieved in a commercial search engine create their own particular material reality. Ranking is itself information that also reflects the political, social, and cultural values of the society that search engines operate within..."

Noble S. U. (2018). Algorithms of oppression : how search engines reinforce racism. New York University Press.

Search engines do not exist outside of society, they are a product of our society and, therefore, they are susceptible to all the same biases and cultural values of the society in which they exist. Search engines are not neutral. And neither is generative AI.

 

---

Using Generative AI

In a letter to the editor of Accountability in Research, Haman and Školník (2023) outlined issues they encountered when using ChatGPT for literature searching:

We used ChatGPT on 23 February 2023. ChatGPT was given the prompt: “‘List 10 seminal academic articles in the field of medicine and provide the DOI.” We used this prompt five times, resulting in five unique chats. ChatGPT provided us with 10 publications each time, for 48 unique publications from the five attempts. We found that only 8 of the 50 provided DOIs exist and are listed for the correct publication. At the same time, we searched for the title of the paper itself with the authors and the correct academic publication (even though the DOI may have been wrong), and only in 17 out of 50 cases did the article in question exist in the databases (Google Scholar, PubMed, Semantic Scholar). In 66% of the cases, ChatGPT gave us a non-existent paper.

Haman, M., & Školník, M. (2023). Using ChatGPT to conduct a literature review. Accountability in Research, 1–3. https://doi.org/10.1080/08989621.2023.2185514

As with all other resources, it's important to verify the information provided to ensure it's accurate. This is particularly true of generative AI tools as they have a tendency to "create" references that actually don't exist. If you do use ChatGPT or similar to generate suggested articles, there are some crucial steps to take to verify what is suggested.

Request and check the DOI - when entering a prompt, request that the DOI (Digital Object Identifier) for each article is provided. A DOI is a unique identifier for a journal article, and is applied to all articles that are published. When provided with the search results, cross check the DOI using the link resolver at dx.doi.org.

Use a search engine to check the article title - Copy and paste the article title into a search engine to ensure it actually exists. Enclose the title in speech marks to ensure it searches for the specific article title rather than the words in the title (eg "The Role of Media Literacy in the Governance of Fake News: A Pedagogical Approach").

Check the details of the article - Ensure the journal name, authors and other details all match what has been provided.

Read the article - There is no substitute for reading the article itself! Always read the article to ensure it meets the criteria you set out in your query, do not assume that the AI has accurately interpreted your query and the articles it's found.

Ways Generative AI Can Help

Keeping in mind all the limitations and issues with generative AI, there are some ways it can be useful to support you in your literature searching.

Developing A Search Strategy - A tool like goblin.tools can provide you with a structure to help you plan your literature search. Simply enter a prompt like "Create an effective literature searching strategy" and it will create a step-by-step guide to conducting an effective literature search strategy.

Identify Key Databases - If you're not sure where to search to find the articles you need, a prompt on ChatGPT such as "I need to do a search on psychology and information literacy. What would be the best research databases to search?" will provide a list of relevant databases to focus your searching.

Identifying Useful Keywords - If you are struggling to come up with search terms for your research area, you can ask a tool like ChatGPT to suggest some suitable keywords. For example, entering a prompt such as "Provide suggested search terms for a literature search on artificial intelligence, information literacy and disinformation" may be helpful in coming up with terms you'd not considered when developing your search strategy.

---

AI Search Engines

There are a range of generative AI tools out there that can help with the literature searching process, but again, we'd recommend using specialised databases and LibrarySearch as they provide authority and coverage that generative AI lacks.

Please note: we do not provide tutorials on any of the following tools.

Elicit

Elicit is one of the more widely known tools for literature searching. Featuring a free trial version and a paid version ($10 per month at the time of writing), Elicit enables a fairly sophisticated search engine that provides useful summaries of papers and enables users to upload and ask questions about academic papers. Using the Semantic Scholar database, it provides access to 125 million academic papers (but not all academic papers!).

If you are interested in using Elicit, the video below provides some useful tips to get you started.

Consensus

Consensus is another popular AI based search engine that uses the Semantic Scholar database to provide search results. Unlike Elicit, Consensus offers a free model which enables unlimited searches and unlimited AI-powered filters.

The video below provides a brief introduction to the search engine.

 

 

ResearchRabbit

ResearchRabbit is a free "literature mapping" tool (sign-up required) that enables you to explore the research landscape and suggests relevant articles. By adding papers to the engine, it can help connect you to other relevant papers and helps you to explore citations. You can find a brief demo of ResearchRabbit below.