Banning books is bad. Using inaccurate tools like ChatGPT to justify banning books is bad and shows a spectacular misunderstanding of how these tools work.
At a recent AI library integration workshop, attendees were introduced to Canva's "Magic Design" tool, which auto-generates presentations based on a user's topic description, combining draft text and stock images. While not perfect, this feature offers a promising start for those looking for quick presentation drafts, and users can easily customize the generated slides to better fit their needs.
There are a lot of companies working on AI-based products and services right now. In our AI workshops for public librarians, we primarily use tools from OpenAI, the company behind ChatGPT. But we also stay up-to-date on the other players, including Google.
As of today, we still view OpenAI's GPT-4 as the most useful general AI model, whether proprietary or open source. Lots of AI-based products and services, including our own PLAID training app, are built around OpenAI's models. But Google has recently announced a push to integrate their "Duet AI" into their Google Workspace offerings. These are presumably variations of the beta AI tools that have been in Workspace for a while, and they're intended to compete with Microsoft's Copilot offerings in Office.
In a live demo this week, we tried it out!
This brief video (just 2 minutes!) shows off some of the highlights of our "Public Libraries & AI" training that we began earlier this year! Our next sessions begin in September so sign up to learn more about how AI works and how it can work in your library. (closed-captions available)
The folks at LibraryThing recently released Talpa.AI, an AI-powered book search tool that can search metadata - including cover images from LibraryThing's collection.
I thought I'd share some of the links that we're reading in this week's PLAID workshops...
This week's "AI in the News" roundup:
Pew says "few have tried" ChatGPT -- but 1 in 7 doesn't sound too "few" to me. Good article on the spread of ChatGPT in the US:
https://www.pewresearch.org/short-reads/2023/05/24/a-majority-of-america...
In our Library AI workshops, we mention the potential of AI-based tools like ChatGPT to assist with library services like Readers Advisory ("RA"). However, until recently, we only had anecdotal examples of how ChatGPT performed at this task.
So, we've spent some time lately figuring out how to methodically explore ChatGPT's capability as an RA tool. It's slow going -- we've just dipped our toe in the water, so to speak -- but we see some great potential!
Hey, everyone! So let's chat about ChatGPT. It's a handy little tool, right? But it does have a couple of quirks that can be annoying. First off, you may have heard of the "training cutoff". In simple terms, it means ChatGPT knows a bunch of stuff...up until the autumn of 2021. After that? It's kinda like it slept through the past year and a half and missed everything post-2021. Another big bummer: it can't browse the web either, so it's stuck with its 2021 knowledge.
ChatGPT and other AI systems are becoming more popular, but they have drawbacks such as outdated knowledge and a tendency to provide false information, or "hallucinate" in AI terms. Using the "retrieval" technique in AI-based apps can give you the benefits of a chat interface along with more relevant and timely information.
Have you seen one or two articles about Artificial Intelligence (AI) in the news lately? Or maybe one or two thousand? AI is already changing how people work, and that includes library staff. That's why we're kicking off our summer 2023 training season with a deep dive on AI in the library.