(created with Flux Image AI using the words, “A student in a library learning about artficial intelligence.”)
Learning about artificial Intelligence (AI), and how it can be used in libraries, has made me want to dig a bit deeper on the subject by reading and reflecting on this topic. Some of the tools I discovered would be more practical for libraries to use, while others would be more for the sake of fun.
ChatGPT is on version GPT-4, which, “shows sparks of general intelligence (ColdFusion, 2023)”. GPT-4 holds a 25,000 word input and output in answering questions asked of it. It can also recognize images which makes it a multimodal and can give descriptions of what is in the picture (ColdFusion, 2023). Libraries could use AI, such as GPT-4, to help answer patron inquires, although I think this technology would work best with information professional involvement, on more complicated reference transactions. Another solution is that it could answer pre-defined or more basic questions, such as library hours and if there is a material in-stock. Then, if the human component was needed, an information professional could somehow be notified of the question. Of course, policies and processes would need to be created for this relationship.
Another topic of hyperlinked environments that is interesting to look at are realistic AI images. The image above was created by typing words into a box, and clicking “Run”, to create a lifelike image that could be mistaken for a real photo. AI images can be a modern form of artistic espression and AI images could open up a number of programs for libraries to offer users. This includes any related subject, including how to create stunning AI images. Or, as a part of online security, libraries could show people how to detect if an AI image is being used in place of a real photograph. In the creation of AI images, libraries could not only discuss what they are, but also facilitate the best language to input, to get the closest to a desired image. Like the ever-evolving AI, libraries must look to the future and anticipate products and services using these technologies.
Reference
ColdFusion. (2023, March 27). AI is Evolving Faster Than You Think [GPT-4 and beyond]. https://www.youtube.com/watch?v=DIU48QL5Cyk
@acorey you share some very good ideas about dealing with AI in the library setting and how we might educate our users. When I first saw that photo, I thought it was real.
It is scary how realistic the photo that was generated is! As a library worker, trying to balance technology, with caution about technology, can be like a maze.
You almost got me with the photo as well! I looked a little closer and saw that the laptop morphed into the book but otherwise it is very lifelike.
Have you taken Professor Mooradian’s course in AI data and ethics? I am learning a lot and really enjoying the content. We talk a lot about AI and the usage within library spaces and a lot of the things that you discussed are covered in the class as well.
No, I have not taken a class from Professor Mooradian. That sounds like an interesting course! Carlee, thank you for the recommentdation!
@acorey I’m glad that you mentioned AI photos, and even included one in your post! Librarians are already teaching information literacy and how to spot fake news, but now we also have to teach how to spot AI generated media and information as well. AI has become huge, and the generated photos are sometimes extremely hyperrealistic, and it is good to spot the fakes sometimes to not be misled with false information.
@inabookbind I’m sure it is daunting to stay on top of technolgical security, especially for libraries. With the AI generated photos, for example, security experts, and libraries, need to stay one step ahead of the bad guys. For as quickly, and realistic, as AI generated photos are, there are programs and people willing to use it maliciously.
Hi @acorey , how to use AI in libraries is a really hot topic right now! I was at the ALA conference over the summer and saw a presentation by JSTOR. about how they are using AI.
Here are some other articles that may be of interest:
Chen, X. (2023). ChatGPT and its possible impact on library reference services. Internet Reference Services Quarterly, 27(2), 121-129. https://doi.org/10.1080/10875301.2023.2181262
Cox, A. M., & Mazumdar, S. (2022). Defining artificial intelligence for librarians. Journal of Librarianship and Information Science, 0(0), 1-11. https://doi.org/10.1177/09610006221142
Lund, B. D., & Wang, T. (2023, November 3). Chatting about ChatGPT: How may AI and GPT impact academia and libraries? Library Hi Tech News, 26-29. http://dx.doi.org/10.2139/ssrn.4333415
Pence, H. E. (2022). Future of artificial intelligence in libraries. The Reference Librarian, 63(4), 133-143. https://doi.org/10.1080/02763877.2022.2140741
van Dis, E. A. M., Bollen, J., van Rooij, R., Zuidema, W., & Bockting, C. L. (2023, February 9). ChatGPT: Five priorities for research. Nature, 614(7947), 224-226. https://www.nature.com/articles/d41586-023-00288-7
Cheers!
@Arthur Kolat Thank you for the articles! There are so many types of AI, which can be a complex topic. I believe ChatGPT and such are referred to as chatbots. I will be focusing on humanoid robots for the strategic plan I am currently working on.
Sincerely,
AdriAnne
Hi @acorey,
I really enjoyed your topic. When it comes to AI, there’s so much ground to cover and so many perspectives to consider that it quite literally hurts my brain trying to wrap my mind around it. But to this day, I think it is *relatively* possible to distinguish between real and AI-generated content, especially when it comes to visuals.
For instance, I immediately got an eerie feeling that your image was fake or AI-generated, probably due to the fact that I recently made one of my own in Canva, so their similar identifying features is fresh in mind. For me, AI images (on average) tend to be very lifeless and boring, and the face and hair region of yours lacks fine detail which is a dead giveaway.
But to test my distinguishing skills, I did Britannica’s Media Literacy quiz on real or AI images, and I got all of them right, which I strongly believe most people would since we can use contextual cues in addition to visual to decipher between stimuli.
However, when it comes to images of faces alone, it’s absolutely GAME OVER for me. I got 2/10 on the New York Times AI Image Generator Quiz, which was more than embarrassing and very humbling LOL!!! AI’s ability to generate completely new “human” faces has scared me more than any horror film ever could since I learned of the existence of this website, which I loath dearly; it creeps me out so much to know I’m looking into the eyes of someone that isn’t real despite looking real. When it comes to faces alone, I think it’s safe to say the robots have officially won!! Hats off to them ^_^
@Kiana Pouya I agree that it is possibly to tell AI images from real one, eventhough they do look very lifelike! For me, the people in the images look too perfect, like they have had waaay to much plastic surgery. Eerie is a great way to describe it! I’ll take that quiz after I get caught up on some more class work, however, the one image I saw on the website you recommended, of AI generated human faces, is creepy to see! Especially, when you think of scammers and what trouble they could cause with this tool!
@acorey nice touch with the AI featured image! I’ve also seen ChatGPT be recommended as a readers’ advisory tool–something I’ve since tried out and found to be pretty neat (I’ve even asked it to recommend movies for me based on content I like and the mood I’m in). Not that you’d want to rely on AI entirely for that service, but it’s a nice supplement! Also, some librarians recommend utilizing AI for administrative/programming preparation, like asking it to generate a presentation for you or to draft a grant proposal. Of course, some of those generative tasks can lead into murky ownership waters, but if you transform it into your own work or credit your source, I think that helps. Nice post!
@Valancy Thanks! I don’t know quite what to think about chat bots. I’ve only had a little experience with Kingbot (the SJSU library tool) and with shopping, like on Amazon. I agree with you that they should just act as a supplement, not a replacement for human effort.
@acorey that’s fair! The chatbots are only as good as the data going into them, so they are prone to mistakes. I don’t know if you saw the trend a few months back where people asked ChatGPT how many “r’s” are in the word “strawberry,” but that really highlighted how authoritative AI can sound while being 10000% wrong! ChatGPT would insist that the answer was “2 r’s ” until you forced it to count the letters as it spelled the word. I find mistakes like this especially concerning for anyone using AI for medical information, because I think it can be prone to misinforming. (Another example was in AI’s recommendation to eat rocks as part of a healthy diet.) I haven’t used the Kingbot before but should try it out!
I did hear about the r’s in strawberry, with the chatbot. Scary how it could mislead people! Does anyone have a minor (Orwellian?) fear that robots could possibly someday control society?