This recording is from the Future of Museums Summit held October 29–30, 2024. Led by museum professionals Clarissa Buettner, Dorothy Bennett, and Peggy Monahan, the session focused on promoting visitor agency, community, and connection. It explored how cultural institutions can demystify artificial intelligence through the lens of human experience, covering strategies to foster critical thinking about AI rather than avoiding it, addressing biases and ethics in AI-generated content, and equipping both staff and visitors with the skills to understand and engage with emerging technologies.

Transcript

Dorothy Bennett:

Hello, everyone. Welcome to our session on human-centered AI literacy approaches, developing programs for fostering agency, community, and connection. We’re so happy to have you here. We thought we’d take one second for you to introduce yourselves in the chat. And Peg, I’m wondering if you can share our slide.

Peggy Monahan:

[inaudible 00:01:01]. Hi, Amy. Can we share those slides? Oh, there we go.

Dorothy Bennett:

Terrific. Next one, please.

So, if you could just share an example of AI you encountered in your daily life that intrigued you or made you pause, that would be great. And also, introduce yourselves. That might take a little time to think about.

Peggy Monahan:

So, why don’t we introduce ourselves?

Dorothy Bennett:

Yes, next slide.

So, I’m Dorothy Bennett. I’m the director of creative pedagogy at the New York Hall of Science. And I’m here with my colleagues Peggy Monahan.

Peggy Monahan:

Hi.

Dorothy Bennett:

Right? Go ahead.

Peggy Monahan:

Hi, I’m Peggy Monahan. I’m the director of content development at OMCA, the Oakland Museum of California. I use she/her pronouns. And my title just basically means I head up the curators.

Clarissa Buettner:

Hello, everyone. My name is Clarissa Buettner and I’m a research and development specialist at The Tech Interactive. I use she/her pronouns. And I’ll jump in on our question too and say one intriguing thing I’ve heard about AI recently is researchers using AI to attempt to decode whale and dolphin vocalizations, which is just a really cool and interesting example of AI use, though I also saw in the chat, “AI slop in my social feed,” from Faith. Yes, I have also encountered that as well, so two ends of the spectrum.

Dorothy Bennett:

Yes, these are great examples here. AI slop in my social media feed. Oh, isn’t it so true? I definitely experience that one every day.

Well, this is great. We have a short time together, so I think we’re going to move along, next slide, and talk to you a little bit about why we are here, this crew. Just a little backstory, over the past four and a half years, Peggy, Clarissa, myself, and other colleagues in different museums were really beginning to discuss that museums have an important role to play in promoting the public’s awareness of AI. That really centered on how it was touching people’s lives and the essential skills they might need.

Next slide, please.

But much of the conversation at the time was really focused on formal education, and it was techno-centric in nature with a focus on developing learning standards and programs that got people familiar with all the different types of AI that you see represented here, and that if we can get people to look under the hood and understand what it is, they maybe would develop the proficiency they needed. And while there were some discussion of ethics and the dangers that AI also presented, there was far less discussion of those human-centered skills and practices people would need to deal with the way the Alexis and the slop in their social media feed and other things that were shaping their lives, how to really deal with that.

Next slide.

And then you probably all recall, it feels like forever ago, but in November of 2022 ChatGPT happened, and it exploded onto the scene and entered people’s consciousness in ways that other technologies haven’t. And that consciousness was on full display in the periodicals that were wandering through my house during the pandemic. And just looking at this little montage of examples that I collected over time, you could see that AI was really on the minds of almost every industry and also in popular culture, from accounting to economics to endocrinology and diagnostics to fashion and music. And clearly, a part of our everyday lives.

Next slide, please.

So, what we decided to do is really borrow from the traditions of the human-centered AI design communities, who are really focused on this idea of changing the conversation from not what the technology does, but how does it help us? And I love this Ben Schneiderman quote from his book on human-centered AI design, which is really asking us to think about how can AI amplify, augment, and enhance human abilities, so as to empower people, build their self-efficacy, support creativity, recognize responsibility, and promote social connection? Really important thing.

Next slide.

So, that’s what we actually did. To really focus in this, we wanted to bring our colleagues together in informal learning. And we did this last October at the New York Hall of Science and brought together 45 experts. They were practitioners, students, and researchers who came to our museum for a two-day conference, and they really represented a diverse community, from exhibit designers to community outreach people to some of the top AI researchers in the field developing these tools. And our mission was to really promote these forward-thinking ideas about what informal could spend its time doing to promote this new conversation and new approach to AI literacy.

Next slide, please.

Our colleague, who isn’t here right now but we love and adore, actually posed was at the conference, and I thought it was worth sharing here because it really underlined the spirit of what we were trying to do. She asked our attendees to think about what would AI look like if it loved us, much like the makers of Sesame Street did back in the day when they were thinking about children’s television and thinking about what TV would do if it loved people instead of trying to sell it to people. So, it was really trying to change that frame of reference for all of us.

Next slide.

So, what we did is together as a group through our planning committee and at the conference, we actually developed this human-centered AI literacy framework for informal learning. And it was really focused… And there are many frameworks out there, and our colleagues at Digital Promise, for example, have something very close to this. But I think the important thing here that’s useful is that it really focuses in on agency. In these three buckets, agency in using AI in terms of data literacy, proficiency with tools, understanding what AI is, not only what AI is, but where it shapes your daily living.

Agency in interrogating AI, which really gets into those issues of building awareness of bias in AI and also, though, how to unearth it. How do we pick it out? Understanding why diversity in AI matters in terms of the data, in terms of training. And also, having critical conversations about the role AI should have and should not have in our lives. A museum is a great place to do that.

And then finally, agency in designing with AI, which is what are a lot of our colleagues working in computer science and camps and all of that were focusing on a lot, but we were taking a different tack here because what we really wanted to do is engage people in reimagining how AI can be designed to meet human and community needs, to expand notions of who can contribute and how to AI and why that’s important. You don’t only need the technical people leading these conversations. And then finally, building capacity, of course, to design with AI, but for creative expression as well as solving problems and influencing designs of the future.

So, next slide.

So, our goals today is really to work together and identify promising roles that museums can play in fostering AI literacy and human agency with AI but really focusing DEAI at the center. So, we’re going to brainstorm use cases for your setting, we hope at the end. We’re going to explore strategies to increase engagement for humans, for diverse humans, communities. Identify future directions for research and programs, and really also consider how do we continue to grow and form a community of practice, if that’s interested, amongst us? Really focused on what informal learning can do to promote this.

And without further ado, I’m going to hand it over to Peggy, who’s going to drill down a little bit about why us? Why informal learning? Why does it matter?

Peggy Monahan:

So, I get to talk to you about why informal settings, what we do in our spaces. We know that in our spaces you get to explore the things you want the way you want, but there are more nuances to the kind of learning that can happen in our spaces than that. And we have a sense of what our spaces can do, but I wanted to ground us in some of the qualities and the affordances of those exhibits and programs and the types of experiences they provide so we can be thinking about how to use them and let our visitors really dig into AI.

So, we have degrees of freedom to experience, experiment. There’s all kinds of ways to engage that doesn’t have to be one right answer. And also, we have multi-generational groups happening all the time. Kids and adults work on things together and each bring their thoughts and perspectives into the experience. We have an opportunity to give people access to new tools that they don’t have in their other parts of their life. And when you’re there, you’re also with other people, which leads to conversation, which can help you think things through, but also, you’re with people with whom you can continue conversations later. That’s super important.

We provide spaces for identity, wellbeing, and belonging. We create spaces where people are welcome and know they belong here, or at least we should. And we want people to bring their whole selves to the experiences. We want them to feel safe enough to be receptive to new ideas and to share their own ideas.

And finally, our spaces are fun. We can lean into that, and we can give people room to be creative. So, these are just some of the things that you can think about when you’re thinking about what can we do with AI. We have a lot of resources we can bring to bear.

And I’m going to hand it back to Doro.

Dorothy Bennett:

Thanks, Peggy. Just next, please. Thanks.

So, what we’d like to do next is really share three use cases and tools we explored during our conference just to spark your own thinking about this, and then we’re going to follow that up with breakout groups where you get to think about how to promote human-centered AI literacy in your settings and for particular audiences. And Clarissa will explain what we’re going to have you do then.

Next slide.

The three use cases were generated during these design sprints that we did on the second day of the conference. And really, we invited very diverse groups to come together and brainstorm ways to engage different audiences with AI and promote agency across different informal learning contexts. So, they were thinking about an AI literacy goal, like using or interrogating or designing with AI, and then also a target audience, which I think is the most important we’re going to have you do here. And then a program format. For example, if it’s an afterschool program or an exhibit. But again, we found that the audience was the thing that really grounded these.

So, the three cases we have, where one is really looking at why diversity and data matters with Teachable Machine, I’ll talk about that for a few minutes. Clarissa is going to talk about ChatGPT and prompt engineering through Jeopardy and games. And the third use case, Peggy is going to share on using text-to-image generators for family storytelling and memory-making, which I love, and think is amazing.

Okay, next slide.

So, data. Okay, when you talk about AI literacy, you have to talk about data because AI tools and systems, as you know, ChatGPT, large language models, you’ve heard all the buzzwords, very, very many of them rely on large amounts of data. But what’s difficult is that what the data is and where it comes from is really hidden from end users. And so to really promote human-centered literacy and AI literacy, we really feel users need to be aware of what that data is and have opportunities to interrogate it, at least interrogate the data that the tool is using. And it helps if users have a chance to interact with data that they actually can relate to or that’s personally relevant and that’s meaningful. And you can see images of Dear Data here. If you haven’t seen this book, it is amazing, but the point it really makes is data is in the eye of the beholder, so to speak. We humans have that decision-making power to decide what data is.

Next slide, please.

So, one such use case that puts this into focus is Teachable Machine. It’s a free web-based tool from Google that allows you to enter your own forms of data and then train the system to classify that data. It’s a machine learning model, if you will. And it’s really accessible. The data could be in the form of images, like you hold up an image to the webcam or draw your own and hold it up or even upload images. You can put in audio files as data, or you can even do poses and video poses as data. And as it builds up more points of data that you’re feeding it, it learns what your categories mean.

So, next slide.

So, in our use case at the conference and beyond, arts was focused on international high school students. And we figured that they would come from different regions of the country or in the world where their homes might look different and be quite diverse, and so what we thought might be interesting is to have each student draw a picture of the home they lived in back in their home country and then have it become part of the training data set. And you’re basically asking the system to classify things based on the data you give it that you’re training it.

So, on the left here, you get to define the categories. We have house and not house. You can have many more than two. It doesn’t have to be binary. And then you train it. You train it with examples of what’s not house, and you train it with what’s a house. And then on the right, once you run the model, you get to see your confidence level. It identified this one, for example, as a house at 100% confidence level, which is pretty impressive.

Next slide.

But imagine if you trained a system only on images of houses from the United States. What would happen in the model?

Next slide.

So, what we found out is diversity really matters in data, and this tool allows you to really demonstrate that principle. As you add more diverse data to your data set, you see the differences this makes, and you get to explore the way data diversity affects the model’s accuracy. If you think of the left side of the screen as defining houseness and then look at the right side of the screen, houses could look very different. Again, there’s a lot more complexity to this, but this is a way of starting those conversations. And I think what it helps do is make the case in a very simple way of not only the importance of diverse data sets, but who makes decisions as what counts as data is really important in building these data sets because they’re going to bring different perspectives on what house or whatever data you choose means.

So, that’s our first use case. I’m going to pass it on to Clarissa.

Clarissa Buettner:

Thanks, Dorothy.

So, another AI tool that we explored was large language models like ChatGPT. So, one of our groups was tasked with creating a pop-up that engaged middle schoolers with prompt engineering, and they ended up creating something based around the game Jeopardy. So, answers were provided on the board, but the twist was that the students needed to answer in the form of a question that, if fed into ChatGPT, would get a response that matched the answer on the board.

So, some things that the group had to consider were that when engaging with prompt engineering, it’s important to understand that LLMs are only as good as what you put into it as the user. So, how could we encourage users to use, play with, and interrogate LLMs? And how could we make this experience communal? How could we get users and visitors involved and engaged? And how can we give them agency in this process? So, of course what this group came up with is to simplify and gamify bringing it back to the basics. So, something really unplugged.

Next slide, please.

So, we actually took some of these findings to another conference. And when we presented new groups with this concept of the Jeopardy with ChatGPT, instead of exploring that Jeopardy concept, they were actually much more interested in creating their own games with ChatGPT. That was what really held people’s interest. So, getting involved in the creation process and the interaction with the AI was something fun for each group and it provided a lot of space for collaboration between the people in those groups, of which there were, I think the largest group was at least 10 people trying to do this all at the same time.

So, the groups worked together to create their own games in these super quick 10-minute sprints. And each group had really different thoughts and considerations for how their games should work and how it was incorporated. So, I’ve included some examples here and some correlating board games that people were thinking of when they were creating their ChatGPT games. So, they might’ve fed into ChatGPT, “Let’s play a game. Use these one-word clues to guess the correct word or phrase.” Or “We’re playing a game where you need to provide clues that I can use to guess a word or a phrase.”

So, we actually had one group… And so with this also, each group was able to create their own win conditions. And I don’t know about you, but I can be a little bit competitive with board games and things like that, so there was a lot of celebration when we were working with these games and with these rules that people were making and that win condition was met.

So, one really fun example was one group was trying to get ChatGPT to guess a word or phrase based on one-word clues. And they were going through the process, and they really wanted it to guess University of Wisconsin-Madison. There was some personal connections there, so people were very adamant about ChatGPT being able to guess University of Wisconsin-Madison. So, we’re feeding it in words like cheese, badger, which I was led to believe is the mascot of that university, things like that, but ChatGPT was just not getting it. It would be like cheese, dairy farm. Cheese, I don’t know. It was very confused. It was really not giving us what we were looking for.

And over time, the group as a whole realized, oh, it’s actually just guessing based on the most recent clue, so they had to go back and rework the instructions, add new instructions in to be able to get ChatGPT to participate in the game that they wanted. And they did do that. They came up with a whole new prompt, they corrected ChatGPT, fed in the same clues they fed it before, and when it said University of Wisconsin-Madison, I’m pretty sure the table all the way at the other side of the workshop room could hear the yells of joy of, “Yes, we’ve got it. University of Wisconsin-Madison.”

So, being able to work through this game creation process and troubleshooting it really allowed for a lot of joy and flexibility in how people were playing with AI, but also how they were playing with each other. So, my biggest takeaway from introducing this concept into an entirely new group was that there was more joy and willingness to interrogate ChatGPT during this experience because they were engaging with other people, as opposed to it being a one-on-one interaction with an AI.

Next slide, please.

So, we’re talking about museums too. We’re talking about those informal learning spaces. So, I just want to bring up that, like Peggy said, museums have a really unique opportunity to open up space for discussion and interaction around AI. So, how can we take these types of learnings further and move forward in our own institutions? And I feel like this approach can be applied to both larger, more complex experiences like exhibits, but also to the small and unplugged experience. By bringing it back to basics, we can really open up opportunities for everyone to use, interrogate, and design with AI.

Forgive my paper moving.

So, these photos here are actually some unplugged AI activities that we tested at the San Jose Earthquakes Community Day. So, it was a community day event hosted by our local Major League Soccer team in San Jose, California. So, visitors were able to use just simple chart paper setups and building blocks to engage with AI concepts like visual recognition and AI bias. And it was a really big hit. We had plenty of people coming through and able to engage with the activities and with conversation with staff and with each other.

So, sometimes it’s really good to just start from the basics like this, break down a topic and activity or a technology-to-component parts. And there’s a thrill and excitement around new tech, but also a lot of mystery and uncertainty, so breaking down a topic like AI, which is incredibly broad, into smaller pieces provides an easier entry point for participants to explore different use cases and aspects of AI before fitting them all together and applying them to their own interests and lives. So, these unplugged sessions also provided great opportunities, like I said, for deeper discussion between guests and facilitators, but it also provided opportunities for us to gather user feedback and create better experiences in the future.

Next slide, please.

Like I said, those were some smaller unplugged examples, but in museums like mine, The Tech Interactive in San Jose, we have a lot of larger exhibits and interactive experiences that are based around technology. In these types of experiences, there’s still a lot of opportunity to support collaboration and agency with users. And I would even say that it’s more important than ever to consider that diversity of input in AI results in an overall better experience. And the same is true for how we should approach tech and tech learning overall.

So, I’ll give two really brief examples of how visitor input has sped into some of our AI exhibits. Animaker, which is the exhibit that’s shown in the photo on the left, is an experience where guests build an animal out of Duplo blocks, they scan it to make a 3D model, and that model is sent to Animaker, which is an AI using visual recognition for identification. Guests are able to indicate if Animaker was right or wrong and offer corrections to what it guesses.

And Animaker, I would say is pretty good at identifying animals, but it has a much greater love for giraffes, which Dorothy and Peggy have already heard me talk about before, but it’s actually such a strong love for giraffes that it can result in quite a bit of bias. Animaker will assume that a lot more animals are giraffes than they actually are. Well, how did this happen? This happened because Animaker was actually created with guests from the start. They created the testing data and much of the training data that went into making Animaker work. And guests preferred to create giraffes for a variety of different reasons, whether it was they liked giraffes the best out of the options of animals, giraffes were easier to build out of the Duplo block shapes that we had, or they just saw a lot of other people around them making giraffes, and so they made giraffes too. But because of that, we have a really great talking point about bias and the importance of data because guest data and input actually directly affects the output of Animaker.

For another upcoming exhibit that we have, which is the one shown in the sketch on the right, we were creating an exhibit that utilizes AI tools to create an immersive experience, but we actually had to rethink our initial plan about what we were going to do in this space, the reason being that in the excitement of exploring all these great AI tools that were available to us, we actually lost a little bit of focus of what the guest experience was going to be like and it became a little bit muddled. So, we actually had to take a step back, bringing it back to those basics again, and rethink the role that AI played in this and a lot of the other experiences that we want to work on and start creating with AI.

So, we asked ourselves the question, is it helping or hindering guest engagement and fun? And the big question is, is AI driving the experience or is the user? So, we went back to basics with our design. We tried to create an engaging experience for guests first, and then we found places where AI could add to and aid that experience. So, consider that human experience first and consider how users are engaging with AI. And I’ll say again, is AI driving experience or is the user?

And I will pass it on to Peggy now.

Peggy Monahan:

Hi. Thanks, Clarissa.

So, my use case that I get to talk about is using AI-generated images to support family storytelling. The audiences that we were given to ruminate on and consider when we were creating these things was incredibly important part of, I think, this whole process was really centering a certain group of people and considering AI from their point of view and what they wanted to achieve and how we could give them an AI program that would be interesting to them. The audience that my group was considering was families of recent immigrants, and that was super evocative for us. And so these families we thought both have aspirations for a new life, but they also have memories that they share of their old life. And families are really important, and I started thinking about family storytelling and story-keeping.

And so these images are actually of old memories of mine that I don’t have pictures of. I now have images that represent them. I used an image generation program called DALL.E that is now part of ChatGPT-4 to create these images. One represents when my siblings and I would smash dusty mops against a wall of our shed. It was like we were running at the wall like a cross between a knight with a lance and a clown with a big makeup poof, and we’d watch the dust glitter and the streaks of sunlight. And the other represents what we called the picnic rock. It was a rock in a field that was so big that we used to have picnics on it. And it still had iron wedges hammered into it from a failed attempt to split the rocks so they could clear the field, and that fascinated me as a kid. When I texted these images to my mom, she knew exactly what they represented, which was cool.

I wanted to share what it’s like to create images like this. First, for this version, I pasted in some excerpts of old family stories that my mom wrote out, and then I just asked ChatGPT-4 to create some illustrations. And then next, once I had images, I just asked for specific adjustments to get the images more right. Sometimes it worked and sometimes it really, really did not. And I checked in with my mom as I was doing this. I loved living in these little stories, and it was really fun to imagine my mom as a little kid and my grandmother as a young woman. Even though some of these were stories that I’d heard before, they hadn’t hit my brain in quite that way, so I thought this is just a really cool thing to do with a family member.

Now, I haven’t tried this as part of a real museum program, just with my family and with these colleagues at this pre-conference workshop that we did last month at the Association of Science and Technology Centers Conference. But I’m sure that when and if I do try this with visitors, it will need iteration. But here’s why I think this has real potential. I love the way it encourages conversation. It builds on this family dynamic and encourages telling and retelling of these closely held and important stories. These experiences aren’t cold. They can be really emotional, nostalgic, poignant, even funny. For instance, I made some images to go along with a classic family story about when my grandmother swore that she had never farted. That was very funny.

Anyway, this activity keeps you as the expert. These are your memories. You’re in charge. The AI isn’t ever going to be right, so you get to be the expert here. You can work on getting it more right, but the AI is the dummy that’s just helping you out here, so you don’t have to be intimidated by it. They’re your memories and you’re in charge.

And on that note, you get to really explore what AI gets wrong and what it doesn’t understand. For instance, I could not manage to get ChatGPT to make an image with three kids on a classic wooden toboggan like the one on the bottom. I asked for adjustment after adjustment, and it just couldn’t get it. I even uploaded that image on the bottom, hoping that maybe if I showed it what a toboggan looked like, it would put three kids on it. Not even close. At one point, it actually put kids on a sled that was going uphill. It definitely said something about what the data is because, as you guessed, these AI image generators are really good at making normal-looking things and they fall apart when they’re asked for unique specificity, especially when that includes a cultural specificity that doesn’t exist all over its training data.

For instance, a colleague at the ASTC Conference was trying to make an image of her memory of being in a state park with her childhood friend where they were wearing folding chairs as backpacks. I never did get the full story there. I just know this is what she was trying to make an image of. But the first image that showed up was that surreal and disturbing one of the kids standing in the middle of the chairs. But in a way, the one on the right I find even more disturbing because this one came about when she was asking it to make the children into two South Asian girls. And you can see what the AI did. Instantly, it made them teenagers, and it made them sexy, which, ew. Because of course, when you poke more and more, you’re going to see that unspecified people are White and children are very often little boys, and the easy images for these image generators to create are simple remixes of what’s out there.

I had a hard enough time getting it to make this image on the left of my mom visiting my nephew when he was really little. Cole was sleeping on the floor, but he wanted my mom to hold her hand over the bed so that he could hold it while they slept. But it took a lot of adjustments before it would make that and to make something that looked even remotely like. On the plus side, there was one iteration where it gave me a grandmother and grandson pair that wasn’t White, which that was good. I do wish it had given the little boy a blanket or a pillow in that particular image.

I think this idea has a lot of potential for programs that give you an opportunity to use AI and to interrogate what it’s good at and what it isn’t. Of course, even though I like this idea and definitely think it has potential, I recognize a big flaw at the core of it, which is that the people who really need to confront the biases inherent in the standard large language model AIs are the people whose memories would probably be most smoothly represented. Suburban middle or upper class White folks. So, that would be a big tweak we would need to make. But there’s still something about the specificity of people’s funny stories or unique memories that could encourage some real exploration of the limits of what AI can do and what it has to say and what it can’t really say about your experiences.

And with that, I’m going to-

Dorothy Bennett:

Peggy, that’s great. We’re going to pass, ooh, sorry, pass it to Clarissa. But there’s a comment in the chat that I think is a good one, just summing this up, which is, “The question posed earlier is crucial to revisit throughout the machine and design learning process. Is it possible that people in AI collaborate in driving different parts and aspect of the user experience?” And really, it’s asking the question of who’s driving it? And how are we inviting these conversations? And also, the idea of how are we inviting collaboration in very different ways, which I think is good.

Anyway, sorry, I didn’t mean to jump in, but I thought-

Peggy Monahan:

No, that’s totally true. And once you start building more and more, that’s part of the reason why the diversity of people involved in these things matters, is that because there’s so many people that aren’t represented until maybe they could be.

Anyway, so I’m passing this off to Clarissa because we really do want to give you guys a chance to talk about this too.

Clarissa Buettner:

All right. That was definitely a lot of stuff said in a short space of time. So, as Peggy mentioned, we do want to give you guys an opportunity to have a short breakout discussion about ways you might center the human experience in your own work. I also want to encourage everybody, because we will have a little bit of time for Q&A at the end, if you do have any questions that you have for us, feel free to throw them into the Q&A section in the little task bar there.

Back to the breakout room discussions. So, to provide a little bit more framework for you guys for the discussion, next slide, please, we’re going to have everybody break out into breakout rooms of about six people. And each breakout room is going to have a audience that we’d like you guys to focus your discussion on from this list. For example, if you are in breakout room five, you’d be discussing how to engage intergenerational groups or intergenerational families. If we do end up flipping over into the teens, you’d just use the last number of your breakout groups. So, for example, if I were in group 12, I would be discussing grades three through five listeners. Of course, we want the discussion to flow and be helpful to you, so if your group does feel drawn to a particular audience, we’re not going to stop you from discussing what feels most relevant to you.

Next slide, please.

We also have four questions to guide your discussion. So, these questions and the audience are also posted in a resource. That is also in that sidebar. If you look at the little icons on the right next to the chat, it’s the one at the very bottom for resources. And [inaudible 00:38:10].

Peggy Monahan:

I also added a link.

Clarissa Buettner:

Yes.

Peggy Monahan:

I also added a link in the chat. Sorry.

Clarissa Buettner:

Wonderful, thank you. And Peggy’s also posted a link in the chat.

So, we’re going to give everybody about five minutes. Sorry, not five minutes. That’s way too short. About 10 minutes or so to participate in this breakout room discussion, and then we’re going to bring everybody back for a Q&A. Remember to add those in the Q&A portion, if you have them. And share some final takeaways and close out the session. So, Amy, if you wouldn’t mind sending folks-

Welcome back, everybody. Sorry to cut the discussion short, but we’ve got some pretty quick sessions going on today, so I just wanted to bring everybody back to make a few small announcements. And of course, thank you all for joining us for the session today.

Peggy, are you able to bring our slides back up?

Peggy Monahan:

I think so. Oh wait, hold on. Sorry. I had had it very helpfully chosen before, and now I will… It went away.

Clarissa Buettner:

Yeah. Apologize to-

Peggy Monahan:

There we go.

Clarissa Buettner:

… folks who were alone in breakout rooms. I know there was a little bit of a-

Peggy Monahan:

Yeah, that’s awful.

Clarissa Buettner:

People get randomly assigned, and then we had to move everybody around, so I hope you at least found a person or two to be able to chat with for a bit. But if not, we’re going to be opening up a thread in the event feed about this session where we really do encourage you to keep the conversation going, ask any questions that we maybe didn’t get to today. So, we’ll be getting that set up as soon as we can.

Before we end the session, these slides are also going to be available for you as a resource, and they’re going to include that AI literacy framework that Dorothy mentioned earlier on. And then we’re also going to… It also has a list of some of the emerging principles for human-centered AI that we have been mulling over as we’ve been doing this work.

Once again, this is who we are. And we are also including a QR code to the previous conference where we started this work, where we have a lot of resources and a lot of the learnings that we’ve had come out of this session.

Dorothy Bennett:

Including the framework and other presentations there.

Clarissa Buettner:

Yes. And unfortunately, we didn’t get to the questions for the Q&A, but I will try and snag those and make sure they get answered in that thread that we’re going to be opening in the event feed. So, once again, thank you all so much for joining us and please enjoy the rest of the summit.

Peggy Monahan:

Thank you.

Dorothy Bennett:

Thank you, guys.