JSTOR has introduced an AI Research Tool on its platform that can assist with either searching for content or evaluating and exploring content. Patrons must create or log in to an individual JSTOR account to access the AI Research Tool. The tool works with JSTOR journal articles, book chapters, and research reports found, but does not yet support images, audio, video, and text-based primary sources. Interactions with the tool are not used to train LLMs, but JSTOR advises users not to enter personal information into the tool as JSTOR will be reviewing de-identified conversation histories to make product improvements.
Searching for Content
After you sign into your personal account and run a JSTOR search, you will see a toggle at the top of your results that will allow you to switch from 'Keyword results' to 'Semantic Results.' The Semantic Results are designed to be more relevant to natural language queries than the traditional keyword search results. Semantic Results returns the top 25 journal articles, book chapters, or research reports that Temple can access.
Evaluating and Exploring Content
When you access the full-text of a journal article, book chapter, or research report on JSTOR, you will see a toggle in the upper right corner showing the sidebar AI Research Tool is 'On' by default. Switching the AI Research Tool to 'Off' will collapse the sidebar. You can select from JSTOR-provided prompts (What is this about?, Show related content, or Recommended topics) or enter your own free-text question. Answers are drawn from the full text of the document being viewed. Selecting 'Show Related Content' will display links to other JSTOR documents.
