Exploring Research-Focused Generative AI Tools for Libraries and Higher Education

Hello everyone, and thank you so much for joining today's session on research-focused generative AI tools. In this presentation, we'll delve into various types of generative AI, with a particular emphasis on research tools like Consensus, Elicit, and Research Rabbit. We'll also discuss the challenges associated with generative AI and consider how these tools impact instruction and library services.

Types of Generative AI

Generative AI is a rapidly evolving field with a variety of applications. Some of the main types include:

Chatbots: Conversational AI systems like ChatGPT that can generate human-like text responses.
Image Generation and Synthesis Tools: Tools like Midjourney and NightCafe that can create images based on textual prompts.
Research Tools: Our focus today is on research tools such as Consensus, Elicit, and Research Rabbit, which aim to enhance the research process.
Music and Video Generation Tools: AI systems that can compose music or generate videos.
Others: The field is continually expanding, and new tools are being developed as we speak.

Research Generative AI Tools

1. Consensus

Consensus is a search engine that utilizes language models to surface and synthesize insights from academic research papers. According to their website:

"Consensus is not a chatbot, but we use the same technology throughout the product to help make the research process more efficient."

Source Material: The content comes from the Semantic Scholar database, which provides access to a wide range of academic papers.

Mission: Their mission is to use AI to make expert information accessible to all.

Example Usage:

When prompted with the question:

"How do faculty and instructional designers use Universal Design for Learning in higher education?"

Consensus provides a summary at the top of the page, analyzing the top eight papers related to the query. Below the summary, it lists the eight papers, including details like the title, authors, publication venue, and citation count.

Features:

Save, Cite, Share: Users can save articles, generate citations, and share them.
Citation Generation: Similar to many databases, Consensus can generate citations, though users should verify for minor errors.
Study Snapshot: Offers a synthesized overview of a paper's key points and outcomes. Note that generating a snapshot may require AI credits.

AI Credits and Premium Features:

AI Credits: Users have a monthly limit of 20 AI credits in the free version, which are used for premium features like generating study snapshots.
Premium Version: Offers additional features beyond the free version.

2. Elicit

Elicit is a research assistant that uses language models like GPT-3 to automate parts of the research workflow, especially literature reviews.

Functionality:

When asked a question, Elicit shows relevant papers and summarizes key information in an easy-to-use table.

Example Usage:

With the prompt:

"How should generative AI be used in libraries and higher education?"

Elicit provides a summary of the top four papers, including in-text citations that link to the sources.

Features:

Paper Details: Includes paper information, citations, abstract summaries, and main findings.
Additional Columns: Users can add more columns to the results table to customize the information displayed.

Source Material:

Elicit pulls content from Semantic Scholar, searching over 175 million papers.

3. Research Rabbit

Research Rabbit is a research platform that enables users to discover and visualize relevant literature and scholars.

Mission:

To empower researchers with powerful technologies.

Features:

Visualization: Provides visual representations of how papers are interconnected.
Explore Options: Users can explore similar work, earlier work, later work, and linked content.
Authors: Allows exploration of authors and suggested authors in the field.
Export Papers: Users can export lists of papers for further use.

Example Usage:

Starting with one or more articles, users can find similar articles, explore cited works, or see which papers cite the original article. The platform creates a network graph showing the relationships between articles.

Personal Experience:

The presenter found Research Rabbit particularly useful for organizing dissertation literature reviews.

Why Use Generative AI in Libraries?

Generative AI technology is not going away; it's becoming a mainstay in our culture and professional practices. Libraries and librarians need to consider how to respond to this technology.

Supporting Patrons

Should we support patrons in using these new tools or try to prevent them from using them?
It's a balancing act, considering the benefits and challenges.

Advancing Effectiveness and Efficiency

Generative AI tools claim to make research more effective and efficient.
Teaching students how to use and evaluate these tools prepares them for future workplaces where such technologies may be prevalent.

Personal Uses of Generative AI

Making Paragraphs More Concise: Using AI to refine writing.
Rephrasing Assistance: Helping with tricky paraphrasing tasks.
Creating Titles: Generating titles for presentations or programs.
Organizing Articles: Managing literature for dissertations or research projects.
Brainstorming: Generating ideas and exploring new concepts.

Challenges with Generative AI

While generative AI offers many benefits, there are significant challenges to consider.

Privacy and Lack of Transparency

Uncertainty about where these tools get their information and how they process data.
Users may unknowingly input sensitive information.

Quality and Hallucinations

AI can produce inaccurate information or "hallucinations," including ghost sources that don't exist.
Some are beginning to refer to these as "fabrications."

Biases and Blind Spots

AI models can perpetuate biases present in the training data.

Date Range of Content

Some AI tools may have outdated information, as their training data cuts off at a certain point.

Plagiarism and Academic Integrity

Students may misuse AI tools, leading to academic integrity violations.
Detection tools exist but may produce false positives.

Detection Tools and False Positives

Tools designed to detect AI-generated content are not foolproof.

Evaluating Generative AI Tools

The AI ROBOT Test

Developed by Hervol and Wheatley, the AI ROBOT test is a framework for evaluating AI tools, focusing on:

Reliability
Objective
Bias
Ownership
Type

This framework can be used in information literacy instruction to help students and patrons critically assess AI tools. You can read more about it here.

Additional Resources

The presenter has compiled a LibGuide with articles, videos, podcasts, and other resources on generative AI.

Poll Results

In a previous presentation, attendees were polled on their views regarding generative AI.

Should Librarians Embrace Generative AI?

Most respondents believed librarians should either embrace it or respond somewhere in between embracing and avoiding. Only one person suggested that librarians should avoid it.

Which Generative AI Tools Are Potentially Useful for Your Library?

ChatGPT: 134 responses
Elicit: 3 responses
Perplexity: 118 responses
Research Rabbit: 189 responses
NightCafe: 40 responses
Other: 22 responses
Consensus: 103 responses

Upcoming GAL Virtual Conference

The presenter is organizing an upcoming GAL (Generative AI in Libraries) virtual conference titled:

Prompter or Perish: Navigating Generative AI in Libraries

Dates: June 11th, 12th, and 13th

Time: 1 PM to 4 PM Eastern Time

Call for Proposals: Librarians are encouraged to submit proposals and participate in the conference. For more information, visit the conference website.

Contact Information

For further questions or to continue the conversation, you can contact the presenter at:

Email: brienne.dow@briarcliff.edu

Conclusion

Generative AI is a transformative technology with significant implications for libraries and higher education. By understanding and critically engaging with these tools, librarians can better support their patrons and prepare for the future.

Thank you for attending today's session. We look forward to continuing the conversation at the upcoming GAL Virtual Conference.