This will take some digesting but I have to say also that it did also set ideas whirling! I am drawn to exploration about how the machinery behind the experience works. Interrogation, prodding, etc. diminishes mystery, improves fluency!
We hear about the environmental impact of using AI and how each query consumes water and electricity at alarming rates. As librarians we are being taught to teach prompt engineering such that a prompt is efficiently written from the start and minimizes the number of subsequent prompts, to lessen the environmental impact of using AI (LLMs). How do we square this with the power of iteration in searching and prompting toward better results? I'm enamored with Undermind, an AI research application, that first interacts with the user asking a series of questions before it identifies studies (analyzes their methodologies and gaps, etc.) and lists them for the user to go view. The questions it asks cause the user to really think about what information they really want and precisely how they need to receive it. I don't know if being in an app it uses less electricity and water than independent questions within an iterative chat in an LLM. Where is the green light area?
Thank you for the excellent framework, Mike, and the affirmation that AI information literacy is built on a foundation of information literacy. Seems obvious, but it isn't always articulated as such. A question: May I use the image of your 4 Tips in a library guide, with attribution and a link to this post?
Excellent work, Mike. I'm so glad to see the emphasis on iteration and more processing.
This will take some digesting but I have to say also that it did also set ideas whirling! I am drawn to exploration about how the machinery behind the experience works. Interrogation, prodding, etc. diminishes mystery, improves fluency!
We hear about the environmental impact of using AI and how each query consumes water and electricity at alarming rates. As librarians we are being taught to teach prompt engineering such that a prompt is efficiently written from the start and minimizes the number of subsequent prompts, to lessen the environmental impact of using AI (LLMs). How do we square this with the power of iteration in searching and prompting toward better results? I'm enamored with Undermind, an AI research application, that first interacts with the user asking a series of questions before it identifies studies (analyzes their methodologies and gaps, etc.) and lists them for the user to go view. The questions it asks cause the user to really think about what information they really want and precisely how they need to receive it. I don't know if being in an app it uses less electricity and water than independent questions within an iterative chat in an LLM. Where is the green light area?
Thank you for the excellent framework, Mike, and the affirmation that AI information literacy is built on a foundation of information literacy. Seems obvious, but it isn't always articulated as such. A question: May I use the image of your 4 Tips in a library guide, with attribution and a link to this post?
Absolutely.
Really appreciated this advice - thank you for making it freely available.