---
id: "workshop:designing-smarter-engagement-with-ai"
type: "InnovateUS workshop"
title: "Designing Smarter Engagement with AI"
url: "https://innovate-us.org/designing-smarter-engagement-with-ai"
date: "2025-09-30"
duration: "60 minutes"
language: "English"
series: "Reboot Democracy: Designing Democratic Engagement for the AI Era"
instructors: ["Beth Simone Noveck","Dane Gambrell"]
recording: "https://www.youtube.com/watch?v=MUiw8xNxgy4"
transcript: "automatic, from the recording"
source: "InnovateUS (innovate-us.org)"
licence: "CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/), as stated on innovate-us.org"
---

# Designing Smarter Engagement with AI

InnovateUS workshop · 30 September 2025 · 60 minutes

Learn a practical framework for structuring inclusive, AI-supported public engagement. Explore how to define clear goals, reach the right participants, and turn public input into actionable results. Leave with a customizable blueprint for making engagement more strategic, equitable, and outcomes-driven.

## Learning goals

- Learn a practical framework for structuring inclusive, AI-supported public engagement
- Explore how to define clear goals, reach the right participants, and turn public input into actionable results

## Speakers

- **Beth Simone Noveck**, Founder of InnovateUS and Director, Burnes Center for Change and the GovLab
- **Dane Gambrell**, Research Fellow, The GovLab

Part of the series [Reboot Democracy: Designing Democratic Engagement for the AI Era](https://innovate-us.org/workshop-series/democratic-engagement).

Recording: https://www.youtube.com/watch?v=MUiw8xNxgy4

## Transcript

Automatic transcript of the recording: names and terms can be misheard, and the times can be a few seconds off. Each time links to that moment of the video. The transcript alone: https://innovate-us.org/designing-smarter-engagement-with-ai.txt (plain text) and https://innovate-us.org/designing-smarter-engagement-with-ai.json (every segment with its start and end second).

**Dane Gambrell**

[0:00](https://youtu.be/MUiw8xNxgy4?t=0) The results of our poll, while I say that the request to remain mute is only in connection with dogs and cats and kids and ringing phones, we very much want to hear from you throughout and especially at the end because we're going to leave a lot of time for questions. So please feel free to use the chat. We have a lot of experts in the room, people who are doing interesting things in your own work and we are very much here to teach and learn from one another. If you have questions or comments, again, preferably to share those in the Q&A. A button you'll find at the bottom, maybe we can write some instructions in the chat about where to find that Q&A. There's so many buttons at the bottom, might be hard to find. But to use the chat or the Q&A to ask questions and we'll be plenty of time for conversation as part of this. Okay, let me share some slides here. And while I'm doing that, Dane, perhaps you want to quickly introduce yourself and then we'll go ahead and dive in.

[1:04](https://youtu.be/MUiw8xNxgy4?t=64) Sure. Thank you, Beth. My name is Dane Gambrell. I'm a research fellow with the GovLab and with Innovate US where I study collective intelligence and AI and in particular how AI can allow us to reach communities more accessibly and more efficiently and more effectively with these kinds of public engagement that we will talk about in the workshop today. So thank you for the opportunity to participate. Fantastic. Dane has an absolutely encyclopedic knowledge of engagement happening all over the world, what works and what doesn't work. And so I'm delighted that he can join me for this today. So many of you may have heard about the Irish Citizens Assembly, probably one of the most successful in many ways, instances of democratic participation, when a representative sample of 99 members of the Irish public have come together to deliberate, leading to changes in Irish law around abortion, for example, or same-sex marriage. It's what led to Ireland's creating the first successful referendum on same-sex marriage. But even this extraordinarily successful example of engagement has only happened six times in 15 years. In Singapore, they similarly adopted a kind of citizen jury, citizen assembly model, tried it four times, and then stopped. That is a very, very common problem. Citizen engagement, despite having fantastic technology, even before AI, we've had, you know, 30 years of the World Wide Web allowing us to do new forms of engagement, and yet citizen engagement too often remains tokenistic, it remains occasional, or it remains very small. So part of what we're here to do today, our goal for this whole series, is to try to accelerate, improve, and institutionalize what we do by way of citizen engagement. The argument

[3:06](https://youtu.be/MUiw8xNxgy4?t=186) underlying the series that we're running here, together with the Allen Lab for Democratic Renovation, Danielle Allen and her team from Harvard, and we're very grateful for their participation, we are taking turns on running these sessions, is to really figure out, as Danielle talked about in our first session, ways that we can connect participation to decision making and to action. So the conversation we're going to have today is really rooted in some work that Dane and I did a few years ago together with Nesta. Nesta, as many of you know, is the innovation lab of the UK, and together with Nesta, we looked at 30 global case studies of successful forms of citizen engagement and participation, where that engagement was connected to impact, it wasn't just people talking, it was actually institutions listening, and it was forms of engagement that have been adopted, repeated over time. So what we're interested in is what works, and that's what led us to developing to essentially reverse engineering this checklist of nine questions that every innovator that we spoke to, that every public engagement professional that we spoke to, said were really the ingredients, questions they had to be able to answer in order to run a successful engagement. We're here to explore not just what those factors are, but how AI can help us accelerate what we do in connection with that work. Now to be very clear at the outset, we are not talking here about getting faster by replacing engagement with AI. We are not talking about saying, look, you can just go ask chat GPT instead of asking communities. We assume, as we've talked about in our other sessions, that we're all in agreement about the need to do engagement and why it's valuable. So we won't spend a lot of time rehashing that argument. Rather, we're going to

[5:06](https://youtu.be/MUiw8xNxgy4?t=306) get into more nitty-gritty today of really trying to look at these nine factors. The slides are shared, everything we talk about here is shared, and we will put it up online, so no need to take any notes. But as we're going along, if you've had experience, if you agree with these questions, disagree, have figured out good ways of using AI to answer these questions, we'd love for you to share your experience in the chat. Okay, let me pause there, and I'm gonna dive in, and all good? Just checking that everybody can hear and see. Fantastic. Okay, step one. This may seem fairly obvious, but I have seen in my experience, more often than not, people going out and doing engagement because they have to, not because they need to. That is to say, there is some sense that we have to go out and do stakeholder consultation, we have to go out and be seen listening to the public, but we're not really clear what we're asking and why we're asking it. So it is extraordinarily important to define what the purpose is for that engagement, and to ensure that everybody is on the same page about that shared purpose. Again, that may be easier said than done. It sounds obvious in principle, but in practice can be quite hard. When we ran our task force on AI for the state of New Jersey, we wanted to go out and make, we needed to go out and make policy for what we would do as the state around AI, to protect workers, to protect the environment, to upskill our residents and our students. But to do that, and really make those policies effectively in a way that was responding to what citizen need was, what we wanted to do was to go out and understand what the problems were that people were facing. So in order to formulate our questions, in order to even go out and ask

[7:08](https://youtu.be/MUiw8xNxgy4?t=428) people what they thought the questions were, we had to get agreement on what the question was that we needed to ask. And again, I'm going to breeze through a lot of these examples, so if anything is unclear about how we did it, the mechanics, please feel free to put notes in the chat or to ask afterwards. So we used an open source AI tool that was free called All Our Ideas, developed by Princeton and Citizens Foundation, to go out and ask people. And we used AI really to help us formulate that question. You'll see on the screen there just a very short question, but what we did to drive the entire engagement, what we did with the public, what we did with public servants as we went out and talked to stakeholders, was we used AI to really help us write a compelling goal statement for why we are doing this engagement. And again, I think it's an extraordinarily helpful use of AI that can give us an opportunity to define our goals up front, to ensure that we're aligned on what it is we need to know from people so we can understand what success is.

[8:17](https://youtu.be/MUiw8xNxgy4?t=497) If I'm going to go out and do that, and to really make sure that everybody is on the same page, that is to say our internal team, the public itself about why we're asking, all of our stakeholders are clear on what we want to do, we also have to figure out who it is we need to go ask. And that is not always an easy question. We know that hearing from diverse voices can actually help us to make better policy. When we can go out and ask people, and by this I mean both people with lived experiences and people with what I'll call credentialed or academic expertise, when I can hear from both of those types of experts, I'm going to get useful know-how that can help me better understand both what the problem is and what the solutions are of what I might want to develop. So I develop a targeted outreach plan. Before AI, maybe I have an office of public engagement or stakeholder engagement that says these are the usual suspects you need to go ask.

[9:21](https://youtu.be/MUiw8xNxgy4?t=561) But with AI, I can help uncover and map other people I might want to engage with. So for example, during COVID, we partnered, the state of New Jersey partnered with the Federation of American Scientists to create a network of 600 PhD credentialed scientists to help us crowdsource answers to people's COVID questions. That was a very, very hard job to go out and find those 600 people to participate and to get them engaged. Now with AI, I can do such a better job of actually identifying who those people are and speed up that process of outreach. In Brazil, the Senate of Brazil invites citizens to participate in every legislative hearing by writing questions that they want senators to ask of testifying witnesses. How do they do that? They look at who's participated before, but more importantly, I think, they look at what people's profession is. So if you're registered as a doctor or a nurse, and there's a hearing about a healthcare issue, they're going to reach out to you and say, hey, do you have anything that you want to ask? So this ability, again, to use AI to map and identify who it is I might want to ask, whether expert or public, and then to go out and reach those people gives me an opportunity to find the right people. I'm going to skip ahead over this. It's just another example here, but I want to skip to this example, which I think is an even more compelling one. This is a project that Dana and I are both involved in together with colleagues, where we are building an AI tool together with parents and teachers and educators that helps to summarize and translate a child's individualized education plan, the plan you get when your child has a disability in public school. Again, we are engaging with diverse voices, not just with PhD educators, but with families of kids with disabilities in building

[11:23](https://youtu.be/MUiw8xNxgy4?t=683) this tool. So all that to say, as we're thinking about both whom to reach out to, how to reach out to them, and what to reach out to them about, I think there are opportunities here to think about using AI to help us to do that. Let me stop the share here and pass the slides to Dane, who's going to take us forward with our next couple of factors. Great. Thanks, Beth. Just give me a second and I will pull up the slides where you left off and screen share that. And in the meantime, if people have comments, questions, your own experience, please share things in the chat and we'll keep going. Okay, great. I'm assuming everyone can see my slides just fine. Okay, great. Thanks. So naturally, once you've figured out who are the right communities and the right types of people that you're trying to reach, as Beth described, the next step is to think through how you're going to go out and reach those people. So this is really about thinking through the strategies that you're using to reach people and making sure that those strategies are targeted to the interests and the priorities of the communities that you want to hear from. So that's also about making So that's also about making outreach accessible, not just accessible in terms of technical accessibility, although that's part of this, but also accessibility in terms of the language that you're using, the format, the channel, whether that's in person or through online outreach, through social media, through mailing lists, and that you're really reaching people where they are.

[13:02](https://youtu.be/MUiw8xNxgy4?t=782) And why this is so important is that without really targeted outreach and targeted outreach to diverse communities who are underrepresented in the existing decision-making processes, we risk just hearing from the people whose perspectives are already represented in the decisions that we're making, and then we risk reaching out on learning from perspectives from other people that are just as, if not more, valuable. So one of the really exciting things here is that AI can help us to accelerate that outreach process, to do it faster, and to amplify the amount of outreach that we can do. So I'll talk in the next slide about an example of how an AI-powered chatbot can be a good way of doing this, but there's also ways to use, you know, easily accessible and iterative AI tools like StatuteBT, Claude, Gemini, to create multiple versions of outreach materials that can be tailored to different audiences, and to help with translating those materials into multiple languages with simplifying technical jargon to make sure that the language that you're using to advertise the opportunity to participate is as plain and as accessible as possible.

[14:17](https://youtu.be/MUiw8xNxgy4?t=857) So I'll show an example here. So the city of Bogota, Colombia, has deployed an AI-powered WhatsApp chatbot that they call Chatico to improve civic participation and how they deliver services to residents. So the platform is AI capabilities built in that allow it to understand and to process natural language, to answer people's questions, and to guide residents through engagement processes. So to take an example, the city used this tool last year as part of its participatory budgeting campaign, where people proposed ideas and then voted on projects they want to see implemented in their neighborhoods. In just under two weeks, the campaign collected over a hundred and thirty... from more than 50,000 residents who weighed in on over 2,700 budget proposals. And so that made it the largest digital participatory budgeting process ever recorded in Bogota, with this chatbot handling nearly 90% of those digital votes. So this example is notable, not just for the scale, So 57% of people who voted in this process were women.

[15:22](https://youtu.be/MUiw8xNxgy4?t=922) but for who participated. There was also strong representation from migrants and from people with disabilities. And these are all groups that are often excluded from civic processes. And so why this works is that the platform's design really prioritizes plain language, mobile-first design being a WhatsApp chatbot. And it's designed in a way that it guides users through the workflow of how to participate in the participatory budgeting process and why it matters. And all of those design factors help to reduce barriers to participation and make it more accessible and more impactful. So the next step in our guide is thinking through who is the right owner for your participatory process. So why this is important is that we know that these kinds of projects, you're usually not doing them alone. They're usually delivered by coalitions, by groups of partners. So this part of the checklist is really making sure to think through who's responsible for what components of the project, and especially making sure to assign responsibility for making use of the community's input.

[16:32](https://youtu.be/MUiw8xNxgy4?t=992) And then with AI in the mix, another part of this is really thinking through the tough decisions that project owners have to make about what types of tasks AI is and isn't suitable for. And then when it is used, making sure that it's used in a way that's ethical And as Beth described at the beginning, and that's responsible. responsible use is making sure that AI is on tap, that it's not being seen as a replacement not on top, either for human decision-making or for real engagement with communities. One example of how generative AI could be helpful here is by helping to think through what those roles and responsibilities should be, helping, for example, to write up descriptions of those roles, but understanding that it's really you and your partners that are responsible for making the determination of who should be responsible for what. So the example that I'll give here is a project that's ongoing from the state of California where the state is working together with communities impacted by the Los Angeles wildfires to collaboratively design an action plan for rebuilding.

[17:43](https://youtu.be/MUiw8xNxgy4?t=1063) So during its first phase, communities deliberated online about what the state's priority should be when it comes to recovery and rebuilding. And they collected more than 1300 comments using a deliberation platform called Othello that uses AI to help with analyzing and with synthesizing large amounts of public input. The reason though that I give this example here is around the partnership. So the state's government operations agency and California's office of data and innovation together with the Carnegie Endowment for International Peace are leading the development of this program. They're also partnering with the city and county and counties, the city of Los Angeles and the county of Los Angeles, I should say, together with community organizations to make sure that the initiative can reach community members. And so the reason we give this example here is that this is an initiative that's owned within state government. It's been championed by California's governor, Gavin Newsom. And while this is an ongoing process, we don't know yet what the outcomes will be.

[18:45](https://youtu.be/MUiw8xNxgy4?t=1125) Having responsible owners within the state government makes it much more likely that the community's priorities will translate into implementable programs and policies. Okay, so then step five of our checklist is thinking through what the incentives for participation should be. So how do we make sure that people know why they should participate? So those incentives could be extrinsic motivators. So for example, people who suggest an idea get entered into a drawing to win a prize, or they might be given a stipend in exchange for their time, or they could be intrinsic motivators. So just appeals to people's desire to do good and to fulfill their civic duties. Incentives are also about reducing barriers to participation. So particularly if we're talking about something that's an in-person engagement offering support like childcare, like elder care, like support with transportation, transportation vouchers, what are sometimes called wraparound services can help to reduce some of those barriers.

[20:01](https://youtu.be/MUiw8xNxgy4?t=1201) And so this is a practical consideration, but it's also about showing respect for the time, the expertise and the experience that people are contributing. When it comes to AI, AI can be useful here in a bunch of different ways. I'll give an example in the next slide about how real-time summaries and visualizations can serve as motivators. But even something as simple as using AI to create personalized thank you notes or personalized certificates of participation that help people feel like their contributions are valued is something that can be a useful thing. Be helpful. So the example that I'll give here is from a community in London Derry, Northern Ireland. So each year the Voice Matters Initiative convenes a people's panel, basically a citizens assembly where residents of Derry, Northern Ireland talk together about what their community's future should look like.

[21:08](https://youtu.be/MUiw8xNxgy4?t=1268) And so some of the deliberations in past years have explored mental health in the area and how it could be improved or how to create a better future for young people. What's really interesting is this year they've started experimenting with a platform called Echo developed by the Belgian, developed by the company Dembrin, which uses AI to transcribe the conversation, to identify common themes and to create visualizations based on participants' ideas in real time. And so why this is really transformative because in contrast to past years where these deliberations would happen and then facilitators would have to get together over the series of several weekends and compare notes and pull findings from the deliberations manually, now the AI transcription and summarization allows them to pull those insights in real time and to report them back to the participants at the end of each deliberation and also to create visualizations of participants' ideas like you see on the slide here.

[22:11](https://youtu.be/MUiw8xNxgy4?t=1331) And so that's really an incentive because it helps people to understand that their voices are being heard and that they're being valued. In addition, Voice Matters also provides some of those financial supports. So participants receive a stipend for their time and the initiative also covers related costs like childcare. So this together, this helps people see that their voices matter, reduces barriers, and helps to see why participating is a good and impactful use of their time. Okay, so then the last piece of the checklist that I'll talk about before handing it back to Beth How do you define the tasks? is step number six. So step six is really about thinking through what you're asking participants to contribute and then how you're providing instructions and making sure that those instructions are clear and that they're matched to participants' abilities and their expertise. So why this is especially important in an AI-enabled engagement is if you're gonna bring AI into the mix, it's important to kind of think through how you're gonna use AI, whether that's summarizing or clustering, categorizing, and then to make sure that you're collecting responses in a way that allows you to then do that AI-enabled analysis later.

[23:33](https://youtu.be/MUiw8xNxgy4?t=1413) And then thinking about generative AI tools, something like ChatGPT, Cloud, et cetera, can be really helpful in just helping to define and refine instructions to make tasks clearer and accessible and easier to follow. It also can help you to collect input in new ways. So whereas some people might be really comfortable writing up a written description of their idea or of their comment or their opinion, other people might be more comfortable talking through their idea out loud. And these transcription or speech-to-text capabilities can allow those people to be able to get their input that way. So the example that I'll give here is from Bowling Green, Kentucky. So in this case, local leaders from Warren County, Kentucky invited residents of Bowling Green to share their vision for their community's future as part of their long-term planning process. So they had a really simple and clearly defined task, which is they asked people to respond to an open-ended prompt that simply asked, what do you wanna see your community's future look like over the next 25 years?

[24:41](https://youtu.be/MUiw8xNxgy4?t=1481) And then in this online participation platform, participants were shown each other's submissions. They could either agree or disagree with other suggestions. And so where AI comes in is AI really allowed this initiative to work at scale. So nearly 8,000 participants submitted nearly 4,000 ideas, more than a million votes over the course of a month, a level of participation that would not be feasible in an in-person deliberation. And they used the Poll List platform to facilitate the deliberation. They also used Google Jigsaw's sense-making tool to categorize the ideas, to identify areas of agreement and disagreement, and to summarize the overall findings into a really nifty interactive report, one slide of which you see here. And so why we give this example is by really clearly defining those tasks for input, the organizers enabled both scale and depth. So participants knew exactly what was expected of them, how they were meant to give input, and then also how those AI tools were gonna synthesize the contributions.

[25:47](https://youtu.be/MUiw8xNxgy4?t=1547) And so that allowed the organizers to collect a huge amount of input without losing the original meaning of people's ideas. Okay, and with that, I'll hand it back to you, Beth, to talk us through the rest of the checklist. Fantastic. If you wanna keep going with your slides, I think that would be great. Let me pause for a moment to say two things. One is there's a lot of great questions coming in in the chat. Please keep them coming. We're going to wrap up shortly and give you a chance to ask your questions. We'll do some combination of essentially, we'll call on you if you've already asked a question and we'll also invite you to raise your hand. Second is some of the things that Dane has talked about. All of these platforms are profiled by People Powered who spoke for us last time and who have a great resource just on platforms and which thing does which. But again, we'll share all the information that we've talked about here previously. And the Bowling Green project that you just mentioned, Dane, they are coming up in our series and we'll get to hear from Bowling Green, the city of Bowling Green, and their tech providers about how they did what they did.

[26:59](https://youtu.be/MUiw8xNxgy4?t=1619) Okay, so let's come back here. And I just wanna remind everybody that AI here is just the tool. I wanna just refresh that we're arguing that these are the key questions you need to answer. And AI is just a tool in the toolkit to help you potentially answer these questions faster and easier to allow us to do more engagement. So the last three things we need to do and you can keep going here is to really figure out, essentially, do I have a plan? Do I have clear steps? Do have I project manage this thing? Have I worked backwards from the end to figure out what I need to do with each step? not just at each step of one engagement, But more importantly, have I thought through the whole policymaking process? So one of the things we tend to do now because engagement is so difficult is to only do engagement once. We do it at one snapshot or one point in the process of making a policy or designing a service.

[28:05](https://youtu.be/MUiw8xNxgy4?t=1685) Part of our argument here is that what we can do is to think about how to use engagement, whether it's with members of the public, whether it's with credentialed experts, whether it's with a subset of stakeholders, that there's engagement we can do at every stage of the policymaking process. So that can include defining the problem, coming up with solutions, implementing those solutions or really evaluating what we've done. So we wanna think about both the workflow in terms of the whole policymaking process and then the process of running the engagement. We can keep going here. AI can obviously help us with any of that kind of project management task. As Dane already pointed out, so I won't belabor it, and by the way, we're gonna hear much more about this from the city of Hamburg, which is coming up talking about their project. Gen AI can really help to shrink how long it takes us to do many of these phases. So when I can rapidly summarize and draw insights from one phase, it allows me then to do more forms of engagement instead of having to spend many, many weeks, days, long time just analyzing that one stage, I can repeat, rinse and repeat more often and do what is being talked about in the chat, which is, there's a suggestion in the chat to say, if you're talking about tariffs, if you ask economists, you're gonna get a very different answer than if you're asking frontline workers.

[29:36](https://youtu.be/MUiw8xNxgy4?t=1776) And we wanna be able to do both of these things, so we're wrapping up this project management piece that AI can help with, and a lot of these tasks can really help us do more engagement. Let's keep going. Oh, great example here from Vienna's climate team. They collected ideas from residents for projects to improve sustainability in their neighborhood. They use the GoVocal platform. I confess that I am an advisor to GoVocal. You will be hearing from them, I think, in our next session, together with some of their city partners. And they used AI to categorize more than 1,000 submissions, to detect overlaps in those submissions, to serve as emerging themes, to do the kinds of things you already heard about from Dane that a number of AI tools are allowing people to do that are essentially doing what we used to do before by hand, that allowed city experts to then evaluate the ideas for feasibility, refine them collaboratively with the original idea providers to ensure that there was a really technically rigorous proposal.

[30:46](https://youtu.be/MUiw8xNxgy4?t=1846) Then they were also able to create a jury of randomly selected residents to make a final decision about proposals. So in other words, they were able to do both online and offline and multiple forms of engagement because they had a well-structured multi-phase process. We can keep going here. Last two things. Really, really important to decide before you start how you're going to evaluate the inputs that you receive. We can show the next slide here. You need a way to determine are you looking, is this a popularity contest? Are you looking to see how much people like what you're proposing or dislike it? Or are you looking for something different, namely expertise or know-how? Are you looking for ideas for what to do or are you looking for sentiment? And who's going to make that determination? Coming back to the Vienna example, is it going to be credentialed experts or citizens?

[31:50](https://youtu.be/MUiw8xNxgy4?t=1910) Internal to government or external? There should be clear criteria, not just for your own team, but for the public to understand how will your ideas be rated. We can keep going. Part of the reason to do that is, again, the clearer you tell people what's the purpose of their submission, the better they are going to participate. If they know that what you want is ideas for how to fix something, you're going to get that instead of just being told I like this or I dislike this. Gen AI can really help you do the kind of categorization and bucketing and assessment of input based on those criteria that you give it. I'll give you one last example here and we can go on to the next slide. Oh, I have a next example here. Thank you, Dane. So maybe I'll give you this example here. The Senate, which I mentioned before, I think Brazil's Senate, which has four different ways that citizens can participate in the way that they make laws.

[33:00](https://youtu.be/MUiw8xNxgy4?t=1980) I mentioned earlier to you the idea that citizens can actually write questions that senators ask. Well, they also have a process by which citizens can, residents, I should say, can actually propose legislation, can actually make proposals for laws. And they're using AI now to start to bucket and cluster and group and evaluate those those submissions that they're getting. The other example that I was going to give you and we can go forward is that in the rulemaking process, the federal level at the U.S., we have been inviting citizen comments since 1946. We're specifically, by law, not interested in sentiment. We're not interested in whether you like a regulation or not. What we're looking for in that process is ideas for how to make the regulation better. So getting clear on why you're asking and making that clear to the public and making clear how you will evaluate it and then using AI to help you to do that. OK, last thing, and then we're going to open it up to questions here.

[34:04](https://youtu.be/MUiw8xNxgy4?t=2044) Very much related to this question of evaluation, but very important that the public understand and that you understand the number one reason people participate, the number one incentive, that intrinsic incentive that Dane talked about before for engaging, is the notion that your participation matters, that it's relevant. The worst thing that we can do is to ask people and then not listen to them. It's not to say that we have to take every idea on board just because a resident proposes a piece of legislation to the Senate of Brazil doesn't mean they enact it. People have to understand how the ideas will be used. And to have at least some visibility into what that process is. So that means having some idea for what you're going to do. You can go forward here, I think. Is there another picture? I don't know what's coming next. I'm going to be surprised.

[35:04](https://youtu.be/MUiw8xNxgy4?t=2104) I will give you an example from my own experience when I worked for the Obama administration. One of the things we did before Obama took office was to ask the public for their ideas for what Obama should do in his first hundred days in office. So we asked people. We got one hundred and twenty five thousand submissions. And guess what? And you know what we did with those submissions? Nothing, because we had no idea how we were going to use them and we weren't able We had a nice, you know, media opportunity from it, but we didn't really take to tell the public. advantage of what people told us. And I think in the long run, that depresses trust, that reduces people's trust in government. And it's precisely what we don't want to do. So generative AI can really help us think about new ways of producing outputs, whether it's drafting a report or a memo or a summary of just showing people back what their own contributions were so that there is a sense of that people having been listened to. As well as then using AI to help draft those directions or articulate what it is that will be done. Again, it's not that you have to do what people tell you, but you have to make them feel heard.

[36:28](https://youtu.be/MUiw8xNxgy4?t=2188) Let's keep going here to the end. It's something we did with our New Jersey AI task force that I mentioned before that I started with at the beginning. I think it's really important to recognize that the public engagement that we did didn't just lead to a report, but it led to really the implementation of suggestions that we heard from the public to specific policy commitments and outputs and to building the things that people told us that they wanted. We were able to act on community input. So this case, I think, really helps us to illustrate the principle that we have on the checklist, that institutions have to close the loop with residents. If you're going to do a deliberation, then there needs to be some concrete outcome. We don't want to just talk for the sake of talking. Coming back to what Danielle told us in the first session, we want to connect deliberation to decision making and decision making to action and to use AI to help us communicate that better to the public.

[37:38](https://youtu.be/MUiw8xNxgy4?t=2258) So I think is that our last slide, Dane? Just to refresh on the checklist, what we want to do now is open it up to your comments, questions, et cetera. Again, all of the materials will be shared after the fact, along with the transcript summary and video of the conversation. But now we want to open it up to you. So I think what we're going to do, Dane, I don't know if you.

* * *

Original page: https://innovate-us.org/designing-smarter-engagement-with-ai · Licence: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/), as stated on innovate-us.org
