ACRL Artificial Intelligence (AI) Interest Group

Hi everyone! 

Heather and I met for a quick co-conveners planning session, and we're working on an agenda for our first Discussion Group meeting that we'll be posting soon. We're looking forward to meeting up with any of you who can join us for an online get together -- and no worries if you can't, we'll also be sharing plans for many more to come! One thing we'll be asking about is what kinds of activities you'd like to see for this AI Discussion Group. We'll share some options and will also be excited to hear your ideas and recommendations so as you think of things, please let us know or share via the discussion board. 

One activity we're planning is a Weekly Discussion Prompt! 

To kick us off:

What is something you don't know or are curious to learn more about Generative AI or AI in general?

I'll give it a try:

For me ... One concern I've been pondering since I read this article, "Why You Might Soon Be Paid Like an Uber Driver, Even If You're Not One" via Slate, is that I'm uncertain about how AI tools embedded in office systems, like Teams combined with Outlook Calendar, Outlook Mail, etc. might eventually be used to track work activity on one's laptop or other devices and impact how we work and our work environment. I think many personnel roles in higher education have an expectation of autonomy, as well as types of work (reflection, thinking, talking with people, facilitating outreach events, etc.) -- tasks that mean one is not always logged on to communication systems and not always on one's laptop, and such roles may have some amount of protection from the possibility of compensation based on tracked work, or other implications. However, other roles might not have as much space or autonomy.

One curiosity, I have, as someone very new to using AI tools (have briefly tried out ChatGPT, Microsoft CoPilot, and a few others once or up to a few times during trainings or a couple times on my own) is: Which AI tools work best for analyzing and visualizing a dataset you already have? Such as a spreadsheet of interactions pulled from LibCal with data about  consultations, events, workshops, meetings -- with various people as 'owners' or collaborators for some interactions, with various amounts of detail about the topic and outcome of the interaction, etc. This is something I'd like to explore to help with annual (or other) reporting, at least for identifying some themes in data we already collect.

What's on your mind? 

~Ginny

Hi everyone! 

Heather and I met for a quick co-conveners planning session, and we're working on an agenda for our first Discussion Group meeting that we'll be posting soon. We're looking forward to meeting up with any of you who can join us for an online get together -- and no worries if you can't, we'll also be sharing plans for many more to come! One thing we'll be asking about is what kinds of activities you'd like to see for this AI Discussion Group. We'll share some options and will also be excited to hear your ideas and recommendations so as you think of things, please let us know or share via the discussion board. 

One activity we're planning is a Weekly Discussion Prompt! 

To kick us off:

What is something you don't know or are curious to learn more about Generative AI or AI in general?

I'll give it a try:

For me ... One concern I've been pondering since I read this article, "Why You Might Soon Be Paid Like an Uber Driver, Even If You're Not One" via Slate, is that I'm uncertain about how AI tools embedded in office systems, like Teams combined with Outlook Calendar, Outlook Mail, etc. might eventually be used to track work activity on one's laptop or other devices and impact how we work and our work environment. I think many personnel roles in higher education have an expectation of autonomy, as well as types of work (reflection, thinking, talking with people, facilitating outreach events, etc.) -- tasks that mean one is not always logged on to communication systems and not always on one's laptop, and such roles may have some amount of protection from the possibility of compensation based on tracked work, or other implications. However, other roles might not have as much space or autonomy.

One curiosity, I have, as someone very new to using AI tools (have briefly tried out ChatGPT, Microsoft CoPilot, and a few others once or up to a few times during trainings or a couple times on my own) is: Which AI tools work best for analyzing and visualizing a dataset you already have? Such as a spreadsheet of interactions pulled from LibCal with data about  consultations, events, workshops, meetings -- with various people as 'owners' or collaborators for some interactions, with various amounts of detail about the topic and outcome of the interaction, etc. This is something I'd like to explore to help with annual (or other) reporting, at least for identifying some themes in data we already collect.

What's on your mind? 

~Ginny

What is something you don't know or are curious to learn more about Generative AI or AI in general?

In short, I would like to learn about how others in our field are keeping up with AI.

I have an interest in AI. This has led to me serving on a campus-wide task force and leading a community of practice in our library... and sometimes I feel like folks in my library look to me as an expert because of this (which I am far from!). The pace of advancements in AI is overwhelming for me at times. There are so many (constantly evolving) tools out there! Does anyone have strategies to share? Do you focus on a single LLM? Do you try to insert AI into your workflows? 

Hi everyone! 

Heather and I met for a quick co-conveners planning session, and we're working on an agenda for our first Discussion Group meeting that we'll be posting soon. We're looking forward to meeting up with any of you who can join us for an online get together -- and no worries if you can't, we'll also be sharing plans for many more to come! One thing we'll be asking about is what kinds of activities you'd like to see for this AI Discussion Group. We'll share some options and will also be excited to hear your ideas and recommendations so as you think of things, please let us know or share via the discussion board. 

One activity we're planning is a Weekly Discussion Prompt! 

To kick us off:

What is something you don't know or are curious to learn more about Generative AI or AI in general?

I'll give it a try:

For me ... One concern I've been pondering since I read this article, "Why You Might Soon Be Paid Like an Uber Driver, Even If You're Not One" via Slate, is that I'm uncertain about how AI tools embedded in office systems, like Teams combined with Outlook Calendar, Outlook Mail, etc. might eventually be used to track work activity on one's laptop or other devices and impact how we work and our work environment. I think many personnel roles in higher education have an expectation of autonomy, as well as types of work (reflection, thinking, talking with people, facilitating outreach events, etc.) -- tasks that mean one is not always logged on to communication systems and not always on one's laptop, and such roles may have some amount of protection from the possibility of compensation based on tracked work, or other implications. However, other roles might not have as much space or autonomy.

One curiosity, I have, as someone very new to using AI tools (have briefly tried out ChatGPT, Microsoft CoPilot, and a few others once or up to a few times during trainings or a couple times on my own) is: Which AI tools work best for analyzing and visualizing a dataset you already have? Such as a spreadsheet of interactions pulled from LibCal with data about  consultations, events, workshops, meetings -- with various people as 'owners' or collaborators for some interactions, with various amounts of detail about the topic and outcome of the interaction, etc. This is something I'd like to explore to help with annual (or other) reporting, at least for identifying some themes in data we already collect.

What's on your mind? 

~Ginny

Hi everyone! 

Heather and I met for a quick co-conveners planning session, and we're working on an agenda for our first Discussion Group meeting that we'll be posting soon. We're looking forward to meeting up with any of you who can join us for an online get together -- and no worries if you can't, we'll also be sharing plans for many more to come! One thing we'll be asking about is what kinds of activities you'd like to see for this AI Discussion Group. We'll share some options and will also be excited to hear your ideas and recommendations so as you think of things, please let us know or share via the discussion board. 

One activity we're planning is a Weekly Discussion Prompt! 

To kick us off:

What is something you don't know or are curious to learn more about Generative AI or AI in general?

I'll give it a try:

For me ... One concern I've been pondering since I read this article, "Why You Might Soon Be Paid Like an Uber Driver, Even If You're Not One" via Slate, is that I'm uncertain about how AI tools embedded in office systems, like Teams combined with Outlook Calendar, Outlook Mail, etc. might eventually be used to track work activity on one's laptop or other devices and impact how we work and our work environment. I think many personnel roles in higher education have an expectation of autonomy, as well as types of work (reflection, thinking, talking with people, facilitating outreach events, etc.) -- tasks that mean one is not always logged on to communication systems and not always on one's laptop, and such roles may have some amount of protection from the possibility of compensation based on tracked work, or other implications. However, other roles might not have as much space or autonomy.

One curiosity, I have, as someone very new to using AI tools (have briefly tried out ChatGPT, Microsoft CoPilot, and a few others once or up to a few times during trainings or a couple times on my own) is: Which AI tools work best for analyzing and visualizing a dataset you already have? Such as a spreadsheet of interactions pulled from LibCal with data about  consultations, events, workshops, meetings -- with various people as 'owners' or collaborators for some interactions, with various amounts of detail about the topic and outcome of the interaction, etc. This is something I'd like to explore to help with annual (or other) reporting, at least for identifying some themes in data we already collect.

What's on your mind? 

~Ginny

I focus on 3-4 tools or platforms, and learn those platforms in depth. Then I explore a few more tools here and there. I see no reason to rush to learn every new tool or AI application.

Any emerging tech field will have hundreds of competitors, extensions, and tools. A small handful of these will survive the initial hype. Over time, 2-3 systems will absorb the smaller companies and emerge to have a large market share. Anyone remember the Amstrad CPC? How about the Entex Select-A-Game? These were two of the hundreds of PCs and gaming consoles released in the 1980s, and they didn't make it. I choose a tool I think is likely to be the leader for a given purpose, and if I think it might be likely to lose future market share to a more popular (not necessarily better) competitor, then I'm fine to switch my allegiance. I research how and why a company like Google chooses to enter the market, what their policies and best practices are, and I keep that in mind if I find a reason to switch.

I work with GPT as an operating suite; Canva as a graphic design suite; and Diffit for educational materials, though I'm likely to switch to Khanmigo given Khan Academy's present dominance of the K-12 market. Auxiliary tools I learn include Elicit, Consensus, and Iris because my faculty want to learn to use them. I'm not impressed with Google's operating suite, but I'm willing to see where it's going.

------------------------------
Jennifer Goodland
Reference and Instruction Librarian
San Juan College
She/Her/Hers

Original Message:
Sent: Oct 25, 2024 09:14 AM
From: Virginia Pannabecker
Subject: Kicking off our Weekly Discussion Prompt! -- What is something you don't know or are curious to learn more about Generative AI or AI in general?

Hi everyone! 

Heather and I met for a quick co-conveners planning session, and we're working on an agenda for our first Discussion Group meeting that we'll be posting soon. We're looking forward to meeting up with any of you who can join us for an online get together -- and no worries if you can't, we'll also be sharing plans for many more to come! One thing we'll be asking about is what kinds of activities you'd like to see for this AI Discussion Group. We'll share some options and will also be excited to hear your ideas and recommendations so as you think of things, please let us know or share via the discussion board. 

One activity we're planning is a Weekly Discussion Prompt! 

To kick us off:

What is something you don't know or are curious to learn more about Generative AI or AI in general?

I'll give it a try:

For me ... One concern I've been pondering since I read this article, "Why You Might Soon Be Paid Like an Uber Driver, Even If You're Not One" via Slate, is that I'm uncertain about how AI tools embedded in office systems, like Teams combined with Outlook Calendar, Outlook Mail, etc. might eventually be used to track work activity on one's laptop or other devices and impact how we work and our work environment. I think many personnel roles in higher education have an expectation of autonomy, as well as types of work (reflection, thinking, talking with people, facilitating outreach events, etc.) -- tasks that mean one is not always logged on to communication systems and not always on one's laptop, and such roles may have some amount of protection from the possibility of compensation based on tracked work, or other implications. However, other roles might not have as much space or autonomy.

One curiosity, I have, as someone very new to using AI tools (have briefly tried out ChatGPT, Microsoft CoPilot, and a few others once or up to a few times during trainings or a couple times on my own) is: Which AI tools work best for analyzing and visualizing a dataset you already have? Such as a spreadsheet of interactions pulled from LibCal with data about  consultations, events, workshops, meetings -- with various people as 'owners' or collaborators for some interactions, with various amounts of detail about the topic and outcome of the interaction, etc. This is something I'd like to explore to help with annual (or other) reporting, at least for identifying some themes in data we already collect.

What's on your mind? 

~Ginny

------------------------------
Virginia (Ginny) Pannabecker
Assistant Dean and Director, Research Collaboration and Engagement
University Libraries, Virginia Tech
She/Her/Hers,They/Them/Theirs
vpannabe@vt.edu
------------------------------

That is on point, Jennifer. I remember the Tandy PCs from back in the late 80s and appreciate this analogy. I have been referencing the dot com boom, but your computer analogy works better.

I guess I am still learning where to focus my energy (and looking forward to when the hype settles down a bit). 

Elicit is a good one. I have also been exploring scite here lately. I have almost settled on using that in depth (even though it requires a subscription). I will have to look into Consensus and Iris for the sake of awareness. Those are two I have not encountered yet!

------------------------------
Ryan Spellman
Online Learning Librarian
Northern Kentucky University Steely Library
He/Him/His

Original Message:
Sent: Oct 31, 2024 08:19 PM
From: Jennifer Goodland
Subject: Kicking off our Weekly Discussion Prompt! -- What is something you don't know or are curious to learn more about Generative AI or AI in general?

I focus on 3-4 tools or platforms, and learn those platforms in depth. Then I explore a few more tools here and there. I see no reason to rush to learn every new tool or AI application.

Any emerging tech field will have hundreds of competitors, extensions, and tools. A small handful of these will survive the initial hype. Over time, 2-3 systems will absorb the smaller companies and emerge to have a large market share. Anyone remember the Amstrad CPC? How about the Entex Select-A-Game? These were two of the hundreds of PCs and gaming consoles released in the 1980s, and they didn't make it. I choose a tool I think is likely to be the leader for a given purpose, and if I think it might be likely to lose future market share to a more popular (not necessarily better) competitor, then I'm fine to switch my allegiance. I research how and why a company like Google chooses to enter the market, what their policies and best practices are, and I keep that in mind if I find a reason to switch.

I work with GPT as an operating suite; Canva as a graphic design suite; and Diffit for educational materials, though I'm likely to switch to Khanmigo given Khan Academy's present dominance of the K-12 market. Auxiliary tools I learn include Elicit, Consensus, and Iris because my faculty want to learn to use them. I'm not impressed with Google's operating suite, but I'm willing to see where it's going.

------------------------------
Jennifer Goodland
Reference and Instruction Librarian
San Juan College
She/Her/Hers

Original Message:
Sent: Oct 25, 2024 09:14 AM
From: Virginia Pannabecker
Subject: Kicking off our Weekly Discussion Prompt! -- What is something you don't know or are curious to learn more about Generative AI or AI in general?

Hi everyone! 

Heather and I met for a quick co-conveners planning session, and we're working on an agenda for our first Discussion Group meeting that we'll be posting soon. We're looking forward to meeting up with any of you who can join us for an online get together -- and no worries if you can't, we'll also be sharing plans for many more to come! One thing we'll be asking about is what kinds of activities you'd like to see for this AI Discussion Group. We'll share some options and will also be excited to hear your ideas and recommendations so as you think of things, please let us know or share via the discussion board. 

One activity we're planning is a Weekly Discussion Prompt! 

To kick us off:

What is something you don't know or are curious to learn more about Generative AI or AI in general?

I'll give it a try:

For me ... One concern I've been pondering since I read this article, "Why You Might Soon Be Paid Like an Uber Driver, Even If You're Not One" via Slate, is that I'm uncertain about how AI tools embedded in office systems, like Teams combined with Outlook Calendar, Outlook Mail, etc. might eventually be used to track work activity on one's laptop or other devices and impact how we work and our work environment. I think many personnel roles in higher education have an expectation of autonomy, as well as types of work (reflection, thinking, talking with people, facilitating outreach events, etc.) -- tasks that mean one is not always logged on to communication systems and not always on one's laptop, and such roles may have some amount of protection from the possibility of compensation based on tracked work, or other implications. However, other roles might not have as much space or autonomy.

One curiosity, I have, as someone very new to using AI tools (have briefly tried out ChatGPT, Microsoft CoPilot, and a few others once or up to a few times during trainings or a couple times on my own) is: Which AI tools work best for analyzing and visualizing a dataset you already have? Such as a spreadsheet of interactions pulled from LibCal with data about  consultations, events, workshops, meetings -- with various people as 'owners' or collaborators for some interactions, with various amounts of detail about the topic and outcome of the interaction, etc. This is something I'd like to explore to help with annual (or other) reporting, at least for identifying some themes in data we already collect.

What's on your mind? 

~Ginny

------------------------------
Virginia (Ginny) Pannabecker
Assistant Dean and Director, Research Collaboration and Engagement
University Libraries, Virginia Tech
She/Her/Hers,They/Them/Theirs
vpannabe@vt.edu
------------------------------

Hey, Ryan, Bree, and Jennifer,

tl;dr ... probably / perhaps at least in some cases, I could start using AI tools to help me be more concise. :-D Short version is - so appreciative of the awesome, thoughtful points and experiences shared here about choosing and evaluating AI tools!

I really appreciate all the thoughts, ideas, and approaches you're sharing here! Yes, agreed, very on point, Jennifer! And all, I appreciate the theme the conversation has taken to talk about ways to manage the deluge of AI options, approaches to selecting tools to try, and thoughts on evaluating them. Elicit and Consensus are ones various people at our libraries at Virginia Tech are trying out, among others. One that I saw someone use recently, which I need to look into more, is Scribe, which the company describes as 'a documentation tool.' From what it sounds like, you turn it on / set it to record (or something like that - haven't tried it yet myself), and then it follows you while you go through a demo of something you want to share as part of a training or online course or on a guide, etc. It takes screenshots and provides a text description of your actions, that I imagine you can then edit / select from. It vastly reduces the time it takes to provide a text, screenshot,  based how-to guide. And you can embed guides in your LMS, or get a PDF, etc. However, before I use it, a few things I need to look into are:

• How accessible are the materials it produces? (Including the PDFs, the embedded step by step guides viewed in an LMS like Canvas)
  • Can I easily find this information on their site?
• What are their terms and conditions?
  • This is particularly important at Virginia Tech in order to determine if the terms fit within our 'low risk' category of applications / software, etc. -- can I use it (or use funds if needed to get a paid version) right away and submit it through our low risk review process (by financial and legal areas), or will it need to go through more stringent review (that takes more time)?
  • Also, how comfortable am I with what their terms and conditions allow? 
    • Will I be providing access to all or some or limited information stored on any device I use when using this tool? Will Scribe pull and store that information? Will it share it with other entities?
• For reliability and accuracy, since this tool follows your own steps online, I'm less concerned about that in this case, as long as it does represent the actions I took, things I clicked on, links, etc., and hopefully also allows for me to edit content if there is a mistake or error in the AI's interpretation of actions, selections, etc.
• Personally, I'm curious about whether Scribe provides a statement about the tool's specific climate (energy and water use) impact on their site, beyond what seems to be an increasingly common placeholder statement along the lines of "The carbon footprint of generating this response is estimated to be around 0.1 to 0.3 grams of CO2, which is roughly equivalent to driving about 1.3 to 4 feet in a car that gets 25 miles per gallon. This is a very small impact compared to everyday activities like driving or flying." (The statement I received from MS Copilot once when I asked it to tell me the environmental impact of processing each prompt and response during that logged in session - using this prompt, "After this prompt and after all future prompts that I share, please include data about the environmental impact of my asking this prompt and the analysis / response / result that you share with me." -- I haven't played around with more prompts yet to try to get something more specific for each prompt.
  
• Cost of course would also be a key factor - how much can I use the tool for free, what is the cost if I don't use it for free, does paying for it provide a more secure experience (i.e. does it take or share less of my data if I pay for it)
  • And if all the above are acceptable, do I / does my library have the funds to cover this?

What factors do you all consider when you're thinking about AI tools?

And ... this has now lead me to write the tl;dr at the start of this post ... probably / perhaps at least in some cases, I could start using AI tools to help me be more concise. :-D Short version is - so appreciative of the awesome, thoughtful points and experiences shared here about choosing and evaluating AI tools!

~Ginny

That is on point, Jennifer. I remember the Tandy PCs from back in the late 80s and appreciate this analogy. I have been referencing the dot com boom, but your computer analogy works better.

I guess I am still learning where to focus my energy (and looking forward to when the hype settles down a bit). 

Elicit is a good one. I have also been exploring scite here lately. I have almost settled on using that in depth (even though it requires a subscription). I will have to look into Consensus and Iris for the sake of awareness. Those are two I have not encountered yet!

------------------------------
Ryan Spellman
Online Learning Librarian
Northern Kentucky University Steely Library
He/Him/His

Original Message:
Sent: Oct 31, 2024 08:19 PM
From: Jennifer Goodland
Subject: Kicking off our Weekly Discussion Prompt! -- What is something you don't know or are curious to learn more about Generative AI or AI in general?

I focus on 3-4 tools or platforms, and learn those platforms in depth. Then I explore a few more tools here and there. I see no reason to rush to learn every new tool or AI application.

Any emerging tech field will have hundreds of competitors, extensions, and tools. A small handful of these will survive the initial hype. Over time, 2-3 systems will absorb the smaller companies and emerge to have a large market share. Anyone remember the Amstrad CPC? How about the Entex Select-A-Game? These were two of the hundreds of PCs and gaming consoles released in the 1980s, and they didn't make it. I choose a tool I think is likely to be the leader for a given purpose, and if I think it might be likely to lose future market share to a more popular (not necessarily better) competitor, then I'm fine to switch my allegiance. I research how and why a company like Google chooses to enter the market, what their policies and best practices are, and I keep that in mind if I find a reason to switch.

I work with GPT as an operating suite; Canva as a graphic design suite; and Diffit for educational materials, though I'm likely to switch to Khanmigo given Khan Academy's present dominance of the K-12 market. Auxiliary tools I learn include Elicit, Consensus, and Iris because my faculty want to learn to use them. I'm not impressed with Google's operating suite, but I'm willing to see where it's going.

------------------------------
Jennifer Goodland
Reference and Instruction Librarian
San Juan College
She/Her/Hers

Original Message:
Sent: Oct 25, 2024 09:14 AM
From: Virginia Pannabecker
Subject: Kicking off our Weekly Discussion Prompt! -- What is something you don't know or are curious to learn more about Generative AI or AI in general?

Hi everyone! 

Heather and I met for a quick co-conveners planning session, and we're working on an agenda for our first Discussion Group meeting that we'll be posting soon. We're looking forward to meeting up with any of you who can join us for an online get together -- and no worries if you can't, we'll also be sharing plans for many more to come! One thing we'll be asking about is what kinds of activities you'd like to see for this AI Discussion Group. We'll share some options and will also be excited to hear your ideas and recommendations so as you think of things, please let us know or share via the discussion board. 

One activity we're planning is a Weekly Discussion Prompt! 

To kick us off:

What is something you don't know or are curious to learn more about Generative AI or AI in general?

I'll give it a try:

For me ... One concern I've been pondering since I read this article, "Why You Might Soon Be Paid Like an Uber Driver, Even If You're Not One" via Slate, is that I'm uncertain about how AI tools embedded in office systems, like Teams combined with Outlook Calendar, Outlook Mail, etc. might eventually be used to track work activity on one's laptop or other devices and impact how we work and our work environment. I think many personnel roles in higher education have an expectation of autonomy, as well as types of work (reflection, thinking, talking with people, facilitating outreach events, etc.) -- tasks that mean one is not always logged on to communication systems and not always on one's laptop, and such roles may have some amount of protection from the possibility of compensation based on tracked work, or other implications. However, other roles might not have as much space or autonomy.

One curiosity, I have, as someone very new to using AI tools (have briefly tried out ChatGPT, Microsoft CoPilot, and a few others once or up to a few times during trainings or a couple times on my own) is: Which AI tools work best for analyzing and visualizing a dataset you already have? Such as a spreadsheet of interactions pulled from LibCal with data about  consultations, events, workshops, meetings -- with various people as 'owners' or collaborators for some interactions, with various amounts of detail about the topic and outcome of the interaction, etc. This is something I'd like to explore to help with annual (or other) reporting, at least for identifying some themes in data we already collect.

What's on your mind? 

~Ginny

------------------------------
Virginia (Ginny) Pannabecker
Assistant Dean and Director, Research Collaboration and Engagement
University Libraries, Virginia Tech
She/Her/Hers,They/Them/Theirs
vpannabe@vt.edu
------------------------------

Hey, Ryan, Bree, and Jennifer,

tl;dr ... probably / perhaps at least in some cases, I could start using AI tools to help me be more concise. :-D Short version is - so appreciative of the awesome, thoughtful points and experiences shared here about choosing and evaluating AI tools!

I really appreciate all the thoughts, ideas, and approaches you're sharing here! Yes, agreed, very on point, Jennifer! And all, I appreciate the theme the conversation has taken to talk about ways to manage the deluge of AI options, approaches to selecting tools to try, and thoughts on evaluating them. Elicit and Consensus are ones various people at our libraries at Virginia Tech are trying out, among others. One that I saw someone use recently, which I need to look into more, is Scribe, which the company describes as 'a documentation tool.' From what it sounds like, you turn it on / set it to record (or something like that - haven't tried it yet myself), and then it follows you while you go through a demo of something you want to share as part of a training or online course or on a guide, etc. It takes screenshots and provides a text description of your actions, that I imagine you can then edit / select from. It vastly reduces the time it takes to provide a text, screenshot,  based how-to guide. And you can embed guides in your LMS, or get a PDF, etc. However, before I use it, a few things I need to look into are:

• How accessible are the materials it produces? (Including the PDFs, the embedded step by step guides viewed in an LMS like Canvas)
  • Can I easily find this information on their site?
• What are their terms and conditions?
  • This is particularly important at Virginia Tech in order to determine if the terms fit within our 'low risk' category of applications / software, etc. -- can I use it (or use funds if needed to get a paid version) right away and submit it through our low risk review process (by financial and legal areas), or will it need to go through more stringent review (that takes more time)?
  • Also, how comfortable am I with what their terms and conditions allow? 
    • Will I be providing access to all or some or limited information stored on any device I use when using this tool? Will Scribe pull and store that information? Will it share it with other entities?
• For reliability and accuracy, since this tool follows your own steps online, I'm less concerned about that in this case, as long as it does represent the actions I took, things I clicked on, links, etc., and hopefully also allows for me to edit content if there is a mistake or error in the AI's interpretation of actions, selections, etc.
• Personally, I'm curious about whether Scribe provides a statement about the tool's specific climate (energy and water use) impact on their site, beyond what seems to be an increasingly common placeholder statement along the lines of "The carbon footprint of generating this response is estimated to be around 0.1 to 0.3 grams of CO2, which is roughly equivalent to driving about 1.3 to 4 feet in a car that gets 25 miles per gallon. This is a very small impact compared to everyday activities like driving or flying." (The statement I received from MS Copilot once when I asked it to tell me the environmental impact of processing each prompt and response during that logged in session - using this prompt, "After this prompt and after all future prompts that I share, please include data about the environmental impact of my asking this prompt and the analysis / response / result that you share with me." -- I haven't played around with more prompts yet to try to get something more specific for each prompt.
  
• Cost of course would also be a key factor - how much can I use the tool for free, what is the cost if I don't use it for free, does paying for it provide a more secure experience (i.e. does it take or share less of my data if I pay for it)
  • And if all the above are acceptable, do I / does my library have the funds to cover this?

What factors do you all consider when you're thinking about AI tools?

And ... this has now lead me to write the tl;dr at the start of this post ... probably / perhaps at least in some cases, I could start using AI tools to help me be more concise. :-D Short version is - so appreciative of the awesome, thoughtful points and experiences shared here about choosing and evaluating AI tools!

~Ginny

That is on point, Jennifer. I remember the Tandy PCs from back in the late 80s and appreciate this analogy. I have been referencing the dot com boom, but your computer analogy works better.

I guess I am still learning where to focus my energy (and looking forward to when the hype settles down a bit). 

Elicit is a good one. I have also been exploring scite here lately. I have almost settled on using that in depth (even though it requires a subscription). I will have to look into Consensus and Iris for the sake of awareness. Those are two I have not encountered yet!

------------------------------
Ryan Spellman
Online Learning Librarian
Northern Kentucky University Steely Library
He/Him/His

Original Message:
Sent: Oct 31, 2024 08:19 PM
From: Jennifer Goodland
Subject: Kicking off our Weekly Discussion Prompt! -- What is something you don't know or are curious to learn more about Generative AI or AI in general?

I focus on 3-4 tools or platforms, and learn those platforms in depth. Then I explore a few more tools here and there. I see no reason to rush to learn every new tool or AI application.

Any emerging tech field will have hundreds of competitors, extensions, and tools. A small handful of these will survive the initial hype. Over time, 2-3 systems will absorb the smaller companies and emerge to have a large market share. Anyone remember the Amstrad CPC? How about the Entex Select-A-Game? These were two of the hundreds of PCs and gaming consoles released in the 1980s, and they didn't make it. I choose a tool I think is likely to be the leader for a given purpose, and if I think it might be likely to lose future market share to a more popular (not necessarily better) competitor, then I'm fine to switch my allegiance. I research how and why a company like Google chooses to enter the market, what their policies and best practices are, and I keep that in mind if I find a reason to switch.

I work with GPT as an operating suite; Canva as a graphic design suite; and Diffit for educational materials, though I'm likely to switch to Khanmigo given Khan Academy's present dominance of the K-12 market. Auxiliary tools I learn include Elicit, Consensus, and Iris because my faculty want to learn to use them. I'm not impressed with Google's operating suite, but I'm willing to see where it's going.

------------------------------
Jennifer Goodland
Reference and Instruction Librarian
San Juan College
She/Her/Hers

Original Message:
Sent: Oct 25, 2024 09:14 AM
From: Virginia Pannabecker
Subject: Kicking off our Weekly Discussion Prompt! -- What is something you don't know or are curious to learn more about Generative AI or AI in general?

Hi everyone! 

Heather and I met for a quick co-conveners planning session, and we're working on an agenda for our first Discussion Group meeting that we'll be posting soon. We're looking forward to meeting up with any of you who can join us for an online get together -- and no worries if you can't, we'll also be sharing plans for many more to come! One thing we'll be asking about is what kinds of activities you'd like to see for this AI Discussion Group. We'll share some options and will also be excited to hear your ideas and recommendations so as you think of things, please let us know or share via the discussion board. 

One activity we're planning is a Weekly Discussion Prompt! 

To kick us off:

What is something you don't know or are curious to learn more about Generative AI or AI in general?

I'll give it a try:

For me ... One concern I've been pondering since I read this article, "Why You Might Soon Be Paid Like an Uber Driver, Even If You're Not One" via Slate, is that I'm uncertain about how AI tools embedded in office systems, like Teams combined with Outlook Calendar, Outlook Mail, etc. might eventually be used to track work activity on one's laptop or other devices and impact how we work and our work environment. I think many personnel roles in higher education have an expectation of autonomy, as well as types of work (reflection, thinking, talking with people, facilitating outreach events, etc.) -- tasks that mean one is not always logged on to communication systems and not always on one's laptop, and such roles may have some amount of protection from the possibility of compensation based on tracked work, or other implications. However, other roles might not have as much space or autonomy.

One curiosity, I have, as someone very new to using AI tools (have briefly tried out ChatGPT, Microsoft CoPilot, and a few others once or up to a few times during trainings or a couple times on my own) is: Which AI tools work best for analyzing and visualizing a dataset you already have? Such as a spreadsheet of interactions pulled from LibCal with data about  consultations, events, workshops, meetings -- with various people as 'owners' or collaborators for some interactions, with various amounts of detail about the topic and outcome of the interaction, etc. This is something I'd like to explore to help with annual (or other) reporting, at least for identifying some themes in data we already collect.

What's on your mind? 

~Ginny

------------------------------
Virginia (Ginny) Pannabecker
Assistant Dean and Director, Research Collaboration and Engagement
University Libraries, Virginia Tech
She/Her/Hers,They/Them/Theirs
vpannabe@vt.edu
------------------------------

Hi everyone! 

Heather and I met for a quick co-conveners planning session, and we're working on an agenda for our first Discussion Group meeting that we'll be posting soon. We're looking forward to meeting up with any of you who can join us for an online get together -- and no worries if you can't, we'll also be sharing plans for many more to come! One thing we'll be asking about is what kinds of activities you'd like to see for this AI Discussion Group. We'll share some options and will also be excited to hear your ideas and recommendations so as you think of things, please let us know or share via the discussion board. 

One activity we're planning is a Weekly Discussion Prompt! 

To kick us off:

What is something you don't know or are curious to learn more about Generative AI or AI in general?

I'll give it a try:

For me ... One concern I've been pondering since I read this article, "Why You Might Soon Be Paid Like an Uber Driver, Even If You're Not One" via Slate, is that I'm uncertain about how AI tools embedded in office systems, like Teams combined with Outlook Calendar, Outlook Mail, etc. might eventually be used to track work activity on one's laptop or other devices and impact how we work and our work environment. I think many personnel roles in higher education have an expectation of autonomy, as well as types of work (reflection, thinking, talking with people, facilitating outreach events, etc.) -- tasks that mean one is not always logged on to communication systems and not always on one's laptop, and such roles may have some amount of protection from the possibility of compensation based on tracked work, or other implications. However, other roles might not have as much space or autonomy.

One curiosity, I have, as someone very new to using AI tools (have briefly tried out ChatGPT, Microsoft CoPilot, and a few others once or up to a few times during trainings or a couple times on my own) is: Which AI tools work best for analyzing and visualizing a dataset you already have? Such as a spreadsheet of interactions pulled from LibCal with data about  consultations, events, workshops, meetings -- with various people as 'owners' or collaborators for some interactions, with various amounts of detail about the topic and outcome of the interaction, etc. This is something I'd like to explore to help with annual (or other) reporting, at least for identifying some themes in data we already collect.

What's on your mind? 

~Ginny