Key Takeaways on AI in Libraries from the Charleston Conference

I recently had the opportunity to attend the Charleston Conference, an annual gathering of library leaders and industry partners. The event featured a lot of content on artificial intelligence in the library and information science landscape. It also featured vendors as partners, with most sessions being a combination of a librarian and a vendor, which is not as common in the archives space.

Random Bits & Sound Bites

• "AI has raised the profile of information as a commodity and resource. Librarians are in the middle, and it's our opportunity to usher this technology responsibly."
• "When it comes to AI, you're either a lemming or a luddite."
• "Adding AI is like adding more cowbell. Can we ever even have enough?"
• The phrase "no AI will replace a librarian" is starting to appear in university collective bargaining agreements.
• “Librarians should be willing & able to figure out how to use AI to make libraries better.” (Emily Singley)
• Tech crunch on how bad AI is at math
• “Naive promise between the querier and the tool that what comes back is truth.”
• Metadata & Provenance are important.  (Ben and I have been thinking about how you record provenance from AI-generated derivatives, so it was good to see it brought up here as well.)

The Concept of "De Minimus" and the Library Copyright Alliance

One discussion focused on copyright in the age of generative AI. The Library Copyright Alliance, representing the ALA and ACRL, clarified that current law is sufficient to govern AI-generated content. They emphasized that if an AI-created work is too similar to a copyrighted original, it infringes on that copyright. The alliance also introduced the concept of "de minimis," where minor AI contributions like light editing or image processing don't require disclosure.

The Long Arm of the Law

One of the most compelling sessions was titled "The Long Arm of the Law" and delved into the relationship between copyright and AI. Roy Kaufman from the Copyright Clearance Center advised attendees to closely monitor the high-profile Thomson Reuters v. Ross Intelligence case, as the outcome could set important precedents for the future of AI and copyright law. The discussion also explored the concept of "transformative" use, a right retained by copyright holders, as well as the four-factor test for fair use, which considers who profits from the use of copyrighted material.

Things to Read

• The book Mastering AI: A Survival Guide to Our Superpowered Future by Jeremy Kahn
• Clarivate’s Pulse of the Library 2024 report
• ARL/CNI AI Scenarios: AI-Influenced Futures

Things to Play With

• Khan Academy’s AI powered chatbot, Khanmigo, was recommended.
• I need to play more with NotebookLM from Google. Max Riggsbee recommended this. It was developed with author Steven Johnson, who says: 

The core idea behind the current version of the product is what we call “source-grounded AI”: you define a set of documents that are important to your work—called “sources” in the NotebookLM parlance—and from that point on, you can have an open-ended conversation with the language model where its answers will be “grounded” in the information you’ve selected. It is as if you are giving the AI instant expertise in whatever domain you happen to be working in.

• An interesting experiment using ChatGPT to create a “digital twin” by Daniel Hook:

I want you to assume a new personality that we will create collaboratively. We will call this personality danielintheory and when I ask you to assume this personality you will use the personality that we now create together to answer my questions. The personality will be a digital twin for me. I want you to make 10 statements that will help you understand how this personality should act and respond in a variety of professional scenarios. Each question should be answered by my specifying a number 0 to 9, 0 being that I completely disagree with the statement and that this represents how the personality should act, 9 being that I completely agree with the statement. Stop after each statement to allow me to specify the number that provides the input that you need and then make the next statement.

For general background, danielintheory is the CEO of a software, Al and data company that provides a range of solutions to academics, librarians, research funders, scholarly publishers, governments, corporate research organisations and non-profit organisations. He has a PhD, is an analytical thinker, and has an English sense of humour. While acting as danielintheory you should always be professional and courteous, but you can sometimes inject humour into your responses to lighten the moment.