Applications of generative AI are on the rise, and many of SMU Libraries’ subscription databases are adding these features. A differentiating factor of generative AI tools in databases is that vendors build their own custom tools instead of using third-party ones like ChatGPT, Bard, or Gemini. The vendors’ custom tools rely only on content found within that product. Below are a few examples of how our vendors supplement their products and thoughts from us (a humanities librarian, a social sciences librarian, and a business librarian) on how effective they are.
Scholarly Literature
ProQuest One Literature and many other ProQuest databases offer ProQuest Research Assistant, an AI-based tool, to help with searching and summarizing articles. After entering a search, AI will suggest ways to narrow the query. Unfortunately, these terms seem very broad and don’t seem as helpful as using Subject searching would be. For example, after entering “wallace stevens,” the suggestions for limiting include “modern poetry” and “elizabeth bishop.”
The key takeaway, which summarizes the most salient points from selected articles, is somewhat more helpful. Since articles in the humanities seldom include abstracts, this could be useful in determining relevance to a research topic.
Multiple databases on the EbscoHost platform, such as Academic Search Complete, Business Source Complete, Education Source, and Political Science Complete, all have enabled a new AI tool.
The Generate AI Insights button is visible on the search results page and creates bullet-pointed summaries from one single article.
As with ProQuest One Literature’s tool, this could be helpful for articles that don’t have an abstract or for densely written content where it’s hard to identify the big ideas. However, the summaries can be shallow and shouldn’t be a substitute for reading the full article. Ebsco also says that it will auto-translate the summaries into another language, so if an article is only in French but you are using the platform in English, it will auto-translate it into English. This could be helpful to get an idea about an article’s content before attempting to find a full translation if desired.
JSTOR offers an interactive research tool that provides summaries of highlighted text, shows related research, and suggests similar topics that may be of interest based on the search.
To access the tool, you’ll need to sign up for a personal JSTOR account with your SMU email. Initiate conversations with the tool to gather more article insights, and you can share feedback and rank the usefulness of the responses.
Business
Quite a few business databases are enhancing their products by adding generative AI aspects. The way generative AI is incorporated and its purpose varies across databases. Before you use it, be sure you first understand how it’s being used. Some databases like Pitchbook use it to provide insights into public companies and their transcripts. With each insight, Pitchbook links back to the original source so you can verify the data for yourself.
Other databases like FactSet and Bloomberg intersperse their own proprietary generative AI tools with third-party tools in a variety of ways. Looking at the Online Assistant in FactSet and HELP in Bloomberg can help shed light on methodologies and use cases; however, because both databases partner with outside companies to provide content, it can be difficult sometimes to determine when and where AI is used. Be on the lookout for tools that can be enabled to provide additional insights.
For example, in FactSet, on the Document Search screen, you can select the Transcript Assistant button to help determine themes from earnings calls for public companies.
The New Constructs reports in FactSet are provided by a third-party investment research firm that utilizes its own proprietary AI-driven model. Our subscription only allows us to view the report snapshot. To truly verify the analyst impressions, we would need to look at the underlying data about the company as well as additional external opinions from analysts.
Bloomberg tends to make their AI uses readily transparent with an indicator at the bottom of a screen. Clicking on the Bloomberg AI button makes it easy to learn more about how it’s being incorporated for the specific function at hand. For example, with the popular Financial Analysis (FA) function, Bloomberg uses AI to provide financial insights from both company-provided data and analyst estimates for public companies.
Another use of AI in Bloomberg is their Document Search (DSCO) function, to highlight trends for categories and groups. For example, you can review ESG trends for a public company.
Some databases like Statista use generative AI in the background to help write the content reports but aren’t transparent that they are using it, unless it’s left in by mistake. Note the last bullet that shows an AI model was used to find trends related to the Free ad-supported streaming TV (FAST) market. This may leave you wondering about the quality and human-authorship of other Statista reports.
Regardless of how the database incorporates generative AI or adds enhancements, we recommend applying critical thinking skills to the information before you. Read the original articles, review the abstracts, review the transcripts, and other content to make the most informed decision if what is at hand will be helpful for your research needs. Generative AI can get you started, but the end is up to you.
This post was written by Rebecca Graff, humanities librarian, Naomi Schemm, social sciences librarian, and Melissa Johnson, business librarian. Contact your librarian to learn more about specialized databases for your field or to find the best tools for your research.