Skip to Main Content

AI in Academic Research and Writing: AI Tools for Academic Research & Writing

Tasks that involve searching

For any of these tasks... Free tools Paid tools
  • Finding and summarizing websites that answer your question
     
  • Asking questions or getting a summary of information on a specific website
    Example: Please summarize this: [your link here]

Grounded with web search

Perplexity AI

Microsoft Copilot 

Gemini

Grounded with web search
ChatGPT Plus (GPT 4)

Pro versions of the tools listed in the previous column.

(all of these tools have pro plans for about $20/month)
  • Finding scholarly articles
     
  • Summarizing a particular scholarly article
     
  • Asking questions of a particular scholarly article
     
  • Uploading the PDF of a scholarly article and asking questions or getting a summary.

Start with library databases and Google Scholar. 
Their coverage is more comprehensive than the tools below.

Grounded with databases of scholarly sources
Use these to find more sources that may not have appeared with keyword searching. They use semantic searching, some of them based on Semantic Scholar, others on OpenAlex.

They also include generative AI features, like natural language queries, summarizing, outlining, etc.

These are not 100% free, as most have usage limits.

 Each tool in the previous column also has a paid version.

Wordsmithing tasks (that don't involve search)

Task Free tools Paid tools
  • Brainstorming ideas or examples
     
  • Narrowing your topic ideas for a research paper
     
  • Get ideas for keywords to search in library databases
     
  • Summarizing and outlining information
     
  • Changing the writing level of some text (5 years old, high school, college, faculty level)
     

  • Changing the writing style (make it more humorous, more formal,
    more satirical, more diplomatic, etc.)

Not grounded
 

Claude              

Grounded with web search

ChatGPT 3.5 and 4o

Perplexity AI

Microsoft Copilot

Gemini 

Not grounded

Claude Pro

Grounded with web search

ChatGPT Plus (GPT 4)

Perplexity AI Pro

Microsoft Copilot Pro

Gemini Pro

(all of these tools have pro plans for about $20/month)

Bias

You must ensure that you account for possible biases that your AI model will produce. For instance, the AI model will provide resources that might only feature the most recent sources, have the most peer reviews, or generate unsuccessful resources for your application if the prompt is not clear. You must continually reread the sources provided as they pertain to your specific question; this can lead to biases if you do not search for sources that counter your research question. Generative AI has been known to produce sources that do not exist and create answers that are inaccurate or false. 

Further Readings

Reliability

It is almost impossible to have an AI model that is operating at 100% accuracy. With some AI models, it is found that they can produce false responses, which can then lead to disinformation, skewed research results, and marketing the AI tool as unreputable. It is important to read forums produced by the AI model company that openly recognize that their model will generate inaccurate results or produce bias and how to avoid these issues. 

Further Readings

Prompting

When asking your AI model research questions you will get varied responses depending on how specific you want it to be. Sometimes you must clarify to the model that what it has produced might have not answered; it is important to reword your input to see if this fixes the problem. The AI model will interpret what it thinks you are asking and that is why users will get varied responses. If you ask for the model to answer a question, it should be noted to ask if it can present citations if the application does not already do so. AI models must continually be trained on how to explain its responses by remaining transparent on information so as to remain neutral unless instructed by the user to provide strict evidence on their topic. 

Further Readings

Privacy

Creating an account to use these tools often requires sharing of personal information. The use of these tools will likely also mean sharing information under their terms and conditions of use, and this data may be used to train the models used in a particular tool. You should avoid entering or uploading data or information that you do not have rights to, that are sensitive, or that are restricted by law or license/agreement.

Most, if not all, AI models require an account to use their site even if it is free. Some sites allow you to use preexisting accounts you may already have such as Google, Microsoft, and Apple accounts. The sites you allow access to your preexisting accounts could potentially fail if exposed to cyber security attacks harvesting your personal information. Using a burner account or alias does not provide a barrier to your identity when performing academic research. Using a VPN provides a buffer of security which allows you to stay safe while on the web and downloading resources and is highly recommended for all computers. You can access a free VPN download through OSU (you must be an OSU affiliate in order to use this). 

It is always a good idea to examine the individual privacy and data use policies of each tool and, when appropriate, take measures to pause history or opt out of including your data for training models. For example, see OpenAI's Data Controls FAQ, managing & deleting your Gemini Apps activity and the Gemini Apps Privacy Hub.

For additional readings on privacy and AI, specifically large language model-powered chatbots, see the following:

Attribution

This page was adapted from  AI Literacy in the Age of ChatGPT: Which AI tool for your task? and Ethics of AI for Researchers by University of Arizona Libraries, © 2024 The Arizona Board of Regents on behalf of The University of Arizona, licensed under a Creative Commons Attribution 4.0 International License.

Creative Commons 4.0 license