Digital Scholar Bytes: Resources for Getting Started with AI Platforms

AI has become an exciting technology in today’s rapidly changing world. While AI has yet to replace the traditional library research process, the right tools can greatly enhance the library research experience. However, with new and evolving tools, it can be challenging to know where to start. In this blog post, we will explore the key differences between Generative AI and Retrieval-Augmented Generation (RAG) AI, and highlight a selection of websites, resources, and articles to help you navigate and get started with each approach.

Generative AI vs Research Augmented Generation AI

There are important differences between the AI popularized in the media, such as ChatGPT and the specialized tools utilized by researchers. Generative AI programs like ChatGPT use LLMs or Large Language Models to retrieve information to users’ queries. Once the LLM has run a query, the resulting answer is sent to the researcher. However, there is no guarantee that the answer has the latest information or that the citation is accurate. Additionally, the lack of clarity regarding the sources of information that generative AI bases its answers on, can make results unreliable. That is, citations to articles that don’t exist have been extensively reported.

Picture from “Enrich LLMs with Retrieval Augmented Generation (RAG)”
Picture from “Enrich LLMs with Retrieval Augmented Generation (RAG)”

RAG AI is a partial solution to these problems. RAG AI is different from generative AI in that it is a superset of generative AI. When prompted to provide an answer, RAG models will consult not just the LLM they were trained on but outside data sources. This data source could be either a closed collection of documents or an open collection like the Internet or Semantic Scholar. When the model delivers an answer, the researcher will see the source of information. The research tools that use RAG AI have additional features such as suggesting alternative search phrases and restricting searching to parts of a research paper (e.g. methodology). For a more detailed description of RAG, check out the IBM video titled “What is Retrieval-Augmented Generation?” and the article titled “Enrich LLMs with Retrieval Augmented Generation (RAG).”

Generative AI resources

The Catholic University of America’s page on General Questions about ChatGPT and Artificial Intelligence is a good place to start for Generative AI resources. This page contains basic information about ChatGPT, how it is used, ChatGPT’s shortcomings, and issues related to student academic achievement. The site is updated fairly regularly with information about AI and what you need to consider before incorporating AI into your research strategy. Starting with this page can help you understand how AI can assist in the research process. Familiarizing yourself with the university’s rules will help you avoid issues down the road.  

Linkedin Learning is a great place to find educational resources on generative AI tools. LinkedIn Learning is an online learning platform offered by the university libraries that provides a wide range of training videos on topics such as coding tutorials to time management videos. There is a whole section of videos on AI and prompt engineering. Three courses which might be helpful for beginners are:

These videos should answer the basic questions you might have about AI and provide pointers for how to get started. Many of these LinkedIn Learning videos end by recommending additional videos that cover more advanced AI topics. 

Retrieval Generated (RAG) resources

CUA librarians have created two webinars on RAG AI research platforms. First, the Digital Scholarship workshop, A Review of Generative AI Tools for Research, introduces what RAG AI is and demonstrates how to effectively use several popular AI research platforms, including Perplexity, Semantic Scholar, Scite, and Elicit. Second, two of our librarians gave a presentation on AI research platforms at the annual meeting of the Washington Library Research Consortium in May titled Deciphering the AI Research Platform Maze: A Comparative Analysis. This presentation reported on new and improved features of most of the platforms featured in the first workshop. Also, the workshop introduced a new player in the field, Undermind

Deciphering the AI Research Platform Maze: A Comparative Analysis
Deciphering the AI Research Platform Maze: A Comparative Analysis

If you’re interested in a librarian perspective on AI, Aaron Tay, a Singaporean librarian, offers a comprehensive outlook on AI development through his blog posts. Check out his posts on Retrieval Augmented Generation  and academic search engines or Prompt Engineering with Retrieval Augmented Generation Systems. Tay is generally optimistic about the future of AI in library research but tempers his enthusiasm by addressing current technological limitations such as hallucinations and result accuracy.

Other Libraries’ Libguides

Many libraries have research guides focused on Generative AI and RAG. Below is a list of a few that we’ve found useful. 

These guides cover important topics related to the use of AI in a research context. These guides consider uses of AI, ways in which students and faculty can cite AI tools, and ethical considerations when using AI. Because these pages are updated and curated by academic librarians, you can trust the information to be more reliable than that found on many other websites.

Concluding Thoughts

These resources are not meant to be the final word on your AI journey, but rather a foundation to help you get started. If you do have questions about AI research tools, please reach out to either librarians Kevin Gunn at Gunn@cua.edu or Charles Gallagher at Gallaghercha@cua.edu.    

Charles Gallagher is a Research and Instruction Librarian at The Catholic University of America Libraries

Share this:

Leave a Reply