A mid-April 2026 workshop, "Where does AI fit? Optimizing business development with AI", hosted by the Entrepreneurship & Libraries Conference, began with a panel of librarians looking at some of the issues faced by libraries in their journey towards incorporating AI tools into their work on entrepreneurial topics and with small businesses and startups. Following the discussion was a presentation of an AI toolkit for librarians dealing with entrepreneurial research questions.
Jonathan Torres, Rutgers University, asked "What role should libraries play in helping communities understand and evaluate AI tools?" This is a somewhat tricky question, as the statistics he presented show several gaps in peoples’ understanding of AI. Information sources used to train Large Language Models (LLMs) include Reddit (40%), Wikipedia (26%), YouTube (23%), and Google (23%). These are useful for depicting how humans communicate but lack the content credibility and quality that librarians provide.
Public perception of AI in selected countries in 2025 revealed that 25% had a positive view, but 39% had a negative view. From these statistics, Torres concluded that "Libraries must educate and empower users to confidently use AI". Plus, there’s a digital divide that impacts economic equality—83% of high-income workers use AI, while only 36% of low-income workers do.
Search Skills and Ethics
Turning their attention to skills needed going forward, Torres mentioned becoming better at crafting prompts while Susan Cardner Archambault, Loyola Marymount University, commented that essential librarian skills don’t change. Concentrate on trustworthiness; ask what is missing; and put the tools through their paces to decide on which ones are appropriate for which situation. Try the REACT Framework. She also noted that, although a core principle of using generative AI tools effectively is iteration, students tend not to do that, opting instead for the first answer presented to them.
Consultant Tricia Crutchfield saw some similar behaviors among early stage entrepreneurs, who also tend to simply take results at face value. One example was asking AI to suggest names for a new business, then choosing one without applying any judgement or further research. On the plus side, she thinks AI tools are very beneficial for market research and for competitive intelligence.
HD McKay, Vanderbilt University, shifted the focus to the databases already subscribed to by libraries. The licensing agreements in place often forbid using these for commercial purposes. Determining non-commercial usage can be difficult when considering student use for class assignments, students consulting with an outside small business, a technology transfer department within the university, and open use by walk-in entrepreneurs. The licensing situation could be different for public libraries.
Another cautionary note raised by McKay was the risks involved in trusting AI models when they are overly optimistic. The desire to please, which causes them to become sycophantic, is an intrinsic problem with LLMs. Ethical concerns were addressed by Archambault and Torres. UNC Greensboro has a good LibGuide about using AI for student business research that includes a section on ethics.
The key takeaway from the discussion: Libraries play a crucial role in bridging knowledge gaps, promoting ethical AI use, and empowering communities
AI Toolkit for Librarians
The AI Toolkit for Librarians is the brainchild of Tom Collopy, UNC-Chapel Hill Entrepreneur in Residence. It is essentially a decision support system for responding to inquiries about the many questions that entrepreneurs could have regarding starting their own business. Collopy presented prompting templates that focused on context, such as what kind of business, who is the customer, and what is the problem. He stressed the four elements of prompting—create, critique, push back, then test and refine.
He demonstrated how one source can turn into many assets. He used the Manus tool to create a website, a guide, a book, a podcast, and an audio book based on his toolkit and course data. It was a very quick creative process, bypassing many of the time-consuming activities he would have had doing it manually.
The recording of the workshop is available on the ELC website