Skip to Main Content

Using Generative Artificial Intelligence (Using Gen AI) at MIT

This guide aims to provide guidance for students about using gen AI as part of their studies, as well as general information on Generative AI applications for students and staff, including providing options for various functions.

Don't use Gen AI for Research Searching!

The MIT Library recommends that you do not use Gen AI text applications for research searching.

Why shouldn't Gen AI Text applications be used for research?


For research, text based Gen AI applications (eg Chat GPT, Co-pilot, Bard) give poor results and have limited functionality and transparency.
Text based Gen AI applications are designed to simulate text discussions, and can provide some answers by searching the web, but they are not meant for serious research.

There are a number of reasons why Gen AI apps are not suitable for research:

  • Hallucinatory data - some Gen AI applications will manufacture fictional, made-up references or data. 
  • Results are not reproducible - Gen AI text generators will produce different results each time they are used - but reproducible results are needed for serious/ongoing research.
  • Variable/poor quality results - results often include Wikipedia pages, online books, and free material.
  • You can't search past paywalls - Many quality articles, books and datasets can't be accessed by Gen AI chat apps as they can't get past the paywall. 
  • You can't use filters to control results - Gen AI lack the complex filters that dedicated research databases have, meaning you can't control thinks like currency, source type, language, type of media, etc. 
  • You can't select peer reviewed sources to maintain quality or compliance.
  • Incorrect answers - for whatever reason, some applications are simply incorrect - the answers they provide are wrong or may be the answer to another query.


What should students and researchers use instead?

Students should always use reputable commercial or Open Access sources.
Journal databases, such as MIT's EBSCO, IEEE or Emerald collections, or quality collections of Open Access Journals such as
OpenSpringer, Google Scholar or Proquest Open. For access to our Journal databases see our Journals Libguides  

Students should also use our e Book services, paper books via the catalogue, or our research databases.
 

WMS Login