Skip to Main Content

Gen AI Literature Review Tools at MIT

Using AI tools for Literature Reviews

Tips on writing a query or prompt for the Gen AI app

The effectiveness of the AI prompt can depend on two things:
 

  • How clearly it communicates your intent
  • How well its structured


1. Open Ended Questions
Open-ended questions or ones that require numbers/explanations MAY or MAY NOT work with Gen AI applications. This is a general case with AI.

2. Questions requiring logical reasoning
These can be problematical for Gen AI tools as the applications generally look for and find what has been already written. They can't look at original research, and come to a conclusion themselves. They need to look at other examples of where humans have looked at original research and written about it. The applications will then use that to come to a conclusion.
Because of this, these sorts of questions rarely produce good results. 


3. Tell how you want the output
Generative AI tools can produce output in various ways. Be specific about how you want the output.

4. Use "Do" or  "Don't"
Doing either of these will focus the search results, saving you time in wading through unecessary results.

5. Be specific
Give the AI as much information about the question as possible, try not to be vague:
Bad example: AI, please tell me why we have human conflict?
Good example: AI, please tell me why there was conflict between Israel and Syria in the 6 day war in 1967

6. Think about the format you want the result in
Remember, many different generative AI can produce results for you in different formats. You may need to be specific about exactly what format you want it in. Do you want it in a business report? a Synopsis? A report with graphics? Video? Make sure it is specific 

WMS Login