---
id: "workshop:different-ai-horses-for-different-courses-matching-tools-to-purpose"
type: "InnovateUS workshop"
title: "Different AI Horses for Different Courses: Matching Tools to Purpose"
url: "https://innovate-us.org/different-ai-horses-for-different-courses-matching-tools-to-purpose"
date: "2025-09-24"
duration: "60 minutes"
language: "English"
series: "Reboot Democracy: Designing Democratic Engagement for the AI Era"
instructors: ["Greta Ríos","Nikhil Kumar"]
recording: "https://www.youtube.com/watch?v=2sKTkE6R5OM"
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"
---

# Different AI Horses for Different Courses: Matching Tools to Purpose

InnovateUS workshop · 24 September 2025 · 60 minutes

Explore how to select the right AI tool for the right purpose from deliberation to co-creation, from problem definition to solution identification.

## Learning goals

- Explore how to select the right AI tool for the right purpose from deliberation to co-creation, from problem definition to solution identification

## Speakers

- **Greta Ríos**, Co-Executive Director, People Powered
- **Nikhil Kumar**, Resource Manager, People Powered

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=2sKTkE6R5OM

## 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/different-ai-horses-for-different-courses-matching-tools-to-purpose.txt (plain text) and https://innovate-us.org/different-ai-horses-for-different-courses-matching-tools-to-purpose.json (every segment with its start and end second).

**Greta Ríos**

[0:04](https://youtu.be/2sKTkE6R5OM?t=4) Thank you so much, Agueda. I will start by saying thank you to everyone who's here, but also thank you to the Allen Lab for Democracy Renovation, to the GovLab, and to Innovate U.S. for having us here. I'm going to speak like really quickly about people-powered and then I'm going to give the mic to my colleague, Michio, who's going to talk about the important things today. So what is people-powered and what am I doing here? As Agueda said, my name is Greta and I normally work and live in Mexico City. Right now I am very happy to be in New York at our offices in people-powered. What is people-powered? We are a global hub for participatory democracy and I know that doesn't, that might not ring a bell to many of you. So why people-powered and what is participatory democracy? Well, people-powered is an organization that was formed to address the big global challenges to democracy. We address two big problems. The first one is that most voices are left out of government decisions. When we talk about democracy, people normally think that we have democracy because we can elect who is going to represent us, but that doesn't mean that we really have opportunities to participate in government decision-making. So we think that we need to create ways to make sure that every voice counts because when people participate in decision-making, we have better policies and we have stronger results. But doing this, which in a very technical way is called participatory democracy, is hard. It's hard and if it's not done correctly, it can be even harmful. So we believe that democracy, that participatory democracy, needs to be done well and that's why we have a

[2:14](https://youtu.be/2sKTkE6R5OM?t=134) lot of programs to help advocates do it better. So in a nutshell, people-powered expands, exists because we want to expand people's power to make government decisions by supporting organizations and governments that are building participatory democracy around the world to do it better. So how we do this is we provide support to these governments for them to be able to enable more participatory democracy. We work from below, we have a global community of practice that helps us do this and we have several tiers of involvement that you can interact with us on. We have a lot of communications channels, we are on social media, we create some resources that are there. We have a resource center, we curate the resources that can help you understand what participatory democracy is and how to use it. But we also have support programs where we match you with a mentor or you can take an online course with us and learn how to do this. And most importantly, as I said, we have a community of support. So people-powered is a member-based organization. We have members around the world, around 200 members right now, and we can have organizational members and individual members. So anyone who is working on participatory democracy, either as a researcher, an advocate, maybe someone who is actually in charge of creating participatory programs out there, you can be a people-powered member and meet everyone else in the community and get to our events and share knowledge and also share your concerns and find support in the community. So that's pretty much what we do and today Nikhil is going to talk to you about some of the very concrete work that we have done

[4:17](https://youtu.be/2sKTkE6R5OM?t=257) in relation to AI and other participatory democracy tools. So Nikhil, the floor is all yours.

**Nikhil Kumar**

[4:26](https://youtu.be/2sKTkE6R5OM?t=266) Thank you, Greta. Hi everyone. My name is Nikhil Kumar and I'm based in Brooklyn and as Greta mentioned, I'm the Resource Manager here at People-Powered. And my job is really to manage the development of capacity-building resources to support practitioners like you to plan, implement, and improve participatory processes. And I'm excited to talk today specifically about the guidance that you need to use digital tools and specifically those that incorporate AI features in your work. And I do want to just note before I get started that my background is in policy and when I think about civic tech, I kind of approach it from the perspective of managers in government and civil society who are first focused on supporting good participatory processes that lead to solid policy outcomes. So for today's session, I really want to approach this topic from the perspective of helping you to understand what you need to know, what questions you need to ask when dealing with technology in order to support your work, as opposed to really thinking like, oh AI is really flashy right now, we need to like wedge it into our work. I want to focus today on how we can make the technology work for government and civil society. So why are we here today? The purpose of today's session is to introduce you to the guidance you need to select the right AI tool for the right purpose. And hopefully by the end of today's session, you'll walk away with a framework for matching AI tools to specific goals, real examples of AI supporting deliberation, problem solving, and decision making, the knowledge and tools that you need to evaluate participation platforms, as well as to spot their

[6:33](https://youtu.be/2sKTkE6R5OM?t=393) limitations, the questions to ask before adopting an AI tool, as well as ways to ensure that AI enhances rather than replacing human judgment. So today's agenda, we're actually right on schedule, so that's great. We'll have plenty of time for q&a after the presentation, but we're going to start with a quick warm up. And I'll, I'll be dividing my presentation into three parts, which, you know, this topic is quite broad, but I think hopefully this is going to be a helpful introduction for you. So I'm going to talk about the challenges and opportunities of using AI for participation, as well as the specific uses that we see from global practice before, during, and after a participatory process. And finally, I'll talk about how you can evaluate digital platforms and AI features. So let's just start with a quick warm up. I'd love to do a temperature check in the room. And Agatha, could we launch the first poll? Okay, perfect. So the first question is, have you used digital platforms or tools to support participatory processes before? And no worries if you're quite new to this. Okay, and then I see the second one is also there. So I'm also curious if you've used AI features to support participation as well. Okay, I see some responses coming in.

[8:24](https://youtu.be/2sKTkE6R5OM?t=504) Okay, I think we can close the poll and display the results. Okay, so yeah, this is actually very interesting. I'm seeing kind of an even spread between those of you who have done a lot of digital participation, those of you who have a little bit of experience with it, as well as those with no experience. And then on the other hand, most people, like over two thirds, have not used AI features, or at least you're not aware of it. So I'm curious if you've or at least you're not aware that you've been using AI features to support participation. Because as we'll see, some of them are actually already built into some of the platforms that are used. Okay, great. So that was very helpful. Now I'm going to share a little bit about the challenges and opportunities of using AI for participation.

[9:27](https://youtu.be/2sKTkE6R5OM?t=567) And first, just a little bit of background on why we need guidance on digital participation in AI. So based on the poll, as most of you are aware, there are many different digital participation tools out there, ranging from repurposed tools that people already use every day, like common networking apps like WhatsApp or Facebook, all the way to comprehensive platforms that have been custom made to support participatory processes like participatory budgeting, citizens assemblies, participatory policymaking, and so on. In addition, as you're all experiencing on a daily basis in your life and work, AI is increasingly common. And the participation deliberation space is no exception to this. So many of the platforms that have been used for participation are adopting AI to enhance the experience, both for program hosts and users. And at the same time, given the number of tools and features out there, and the rapid evolution of the technology, it can be difficult for practitioners in government and civil society to find accessible, relevant, and practical guidance. So at People Powered, so at People Powered, working with our partners and our global community, we've developed a guide to digital participation platforms, which we first released in 2022.

[10:52](https://youtu.be/2sKTkE6R5OM?t=652) And this year, actually just last week, we expanded the guidance to include a special section on AI and digital participation. So that's kind of the context of what we're coming to this session with to present to you. The guide is for practitioners in governments, other public institutions, civil society organizations, really any practitioner looking to understand the opportunities and challenges related to digital tools and AI, as well as very practical guidance on how to operationalize those tools in your work. So our process for updating the guide this year was very empirically based. So we worked with the lead writer, Matt Stempeck, from the Civic Tech Field Guide, as well as partners at UNDP and the Friedrich Naumann Foundation. And we surveyed platform developers, as well as users, to understand how people are actually using the tools. So rather than taking a more theoretical approach to what's possible, we're trying to see what's happening on the ground, because that's probably the most useful for our audience. And we used a mix of observation, desk research, interviews, workshops, as well as expert review. And as you can see, the guide is divided into different sections, including an overview of digital participation, a special section on the challenges and opportunities of using AI, guidance for selecting, setting up, and running a digital platform, some thoughts for looking ahead, and additional practical resources.

[12:32](https://youtu.be/2sKTkE6R5OM?t=752) And as I mentioned, we did this work very collaboratively. So we worked with our global member community of participatory democracy practitioners. And we also worked with Matt Stempeck from the Civic Tech Field Guide, who brings a wealth of knowledge on the civic tech landscape, and specifically related to participation. We also worked with the Friedrich Naumann Foundation, based in Taiwan, and with UNDP, to really flesh out the details and the case studies of digital participation that uses AI, so that the guide is really reflective of global practice, and relevant for different audiences. What do we mean by AI? I think AI is a buzzword, obviously, but sometimes it's not really clear exactly what it means. So AI, as we understand it, is really an umbrella term that encompasses an array of different technologies. And basically, right now, we're calling something AI when it allows a computer to act like a human would. And over time, the technology that we take for granted, stops being considered AI. And this is the case, for example, with speech processing.

[13:53](https://youtu.be/2sKTkE6R5OM?t=833) Over decades of work in speech processing and human computer interaction, computers have developed to a point where they can decently understand how we talk, at least some of the time, if the voices are clear, and they're in common languages. And just to kind of frame the discussion today, when we talk about AI, we're including the full subset of methods, ranging from natural language processing, to machine learning. And, of course, a lot of what's currently referred to AI as generative AI, that relies on large language models. For the purpose of our session today, we're really thinking about the actual features that each type of AI unlocks, rather than the underlying computer science. And given how fast moving and controversial AI is, it's really important, and I really want to emphasize this, to approach the issue with a set of principles. So at People Powered, we developed this year an AI policy to guide our work.

[15:00](https://youtu.be/2sKTkE6R5OM?t=900) together with our community. And you can see some of the principles that are included in this policy, including mission alignment, that the AI is human-centered, that it's transparent, that it's accountable, meaning that humans are ultimately responsible for the output and the decisions, even if they use the AI tools. In addition, sustainability, equity and inclusion, learning, and of course privacy and security. So a lot of these principles are aligned with our organizational values, and I think that's really important. And I would also encourage all of you to work with the communities that power your work and the stakeholders that you engage with to develop your own set of principles to guide your approach so that it is grounded in your values and so that you're not just reacting to the latest technology that's being developed. So a few of the concerns related to AI. I would say that some of the concerns are really kind of the counterpoint of all of these principles. So it's possible that the AI might not be transparent, and like it might not be clear how decisions arrived at using AI were calculated, or it might not be accountable. As I said, a human might not ultimately be filtering the output of the AI.

[16:36](https://youtu.be/2sKTkE6R5OM?t=996) So I think, yeah, basically the concerns are that the AI will kind of go against what we value. And then also some specific examples that we've been seeing a lot in research and the media are related to kind of the uncertainty with this new technology. So thinking about what are going to be the economic consequences of the AI? Are there going to be big job losses, including in the governance space? Will it harm people's critical thinking abilities and possibly even kind of take away the human element of participation? I know there's some interesting research now about this. Will the AI hallucinate, plagiarize, be sycophantic? There are ways to mitigate this, such as retrieval augmented generation, but there are no perfect solutions, of course. What is the environmental impact related to water use, energy use, and are the environmental impacts equitable? For example, where data centers are sited is often inequitable. There might be second order effects that we don't even know are going to happen as a result of this new technology.

[17:54](https://youtu.be/2sKTkE6R5OM?t=1074) And then, of course, kind of like more philosophically, the potential existential threat to humanity, which maybe the space isn't really the right place to talk about that, but I think it's also helpful to keep that on our minds. But yeah, it's not all doom and gloom. We are very excited about the potential for AI to improve participation, both for program managers and participants. So one thing that we're seeing on the ground is a boost in efficiency for the running of these processes, which could ultimately lead to more participatory programs, even if participation agencies don't have much budget or staffing, they could use AI to help out. Also, AI is unlocking deeper and new forms of engagement by bringing in new people who don't often engage in kind of innovative methods of engagement. It can reduce digital barriers if designed well. For example, if someone isn't comfortable typing, then they could use speech to share their views, and then the AI can convert that into text, more legible inputs from participants. So that's kind of related to program managers being able to use AI to more quickly and more accurately analyze the inputs, and then also better sense-making of the outcomes of the participatory process, which I'll talk a bit more about.

[19:42](https://youtu.be/2sKTkE6R5OM?t=1182) Okay, so a quick check-in. I'm curious to hear from you guys, what's one challenge that you're concerned about when it comes to AI and participation, as well as one opportunity that you're excited about? So feel free to type in the chat. Ensuring representative samples. Hallucinations, yep. I'm not sure what the next question is. Bias is definitely a challenge. That's definitely something to keep in mind. I'm not sure what the next question is. Privacy. The digital divide, yes. I'm not sure what the next question is. So, I'm not sure what the next question is. Ensuring representative samples. Hallucinations, yep. Also curious about any opportunities that you guys are excited about.

[20:49](https://youtu.be/2sKTkE6R5OM?t=1249) Replacement for face-to-face. That could be an opportunity or a challenge. Broader participation. Definitely, we see that in the case studies. Saving time. Accessibility. Summarizing data. Okay, these are great. Yeah, I think I'm glad to see that you guys are very interested and informed. Okay, so now that we have a shared overall understanding of how AI can fit into participatory processes, I'm going to share some of the specific uses that we're seeing in practice. As I mentioned, these are based on real-world examples that we found in our research with Matt Stempeck, UNDP, the Global People-Powered Community. In the guide, you'll see it can be helpful to think about the uses of AI before, during, and after a participatory process. So, that's how I'm going to break it down here. These are some examples that we've seen in practice of program managers using AI before a participatory process to prepare for the actual process. So, that includes streamlining administrative work, which can often be done with general AI tools like ChatGPT.

[22:12](https://youtu.be/2sKTkE6R5OM?t=1332) Doesn't have to be a specific participatory platform. Training staff to run the process. Swappable AI models is very interesting. It basically provides for greater flexibility in case a new model is developed that is more appealing, maybe more public service-oriented. It could allow to swap that new one in. Forming diverse participant groups, which is often relevant for representative processes like citizens' assemblies and juries, as well as simulating participation to test the platform before you get started. And a couple of more specific examples. We've seen AI being used to plan the process. So, for example, IdeaScale AI provides users with guidance and recommendations to design an innovation campaign. And as you can see in the screenshot, Deliberate also has this feature. Emporando Juntas from Brazil recommends using AI for this, but doesn't yet offer a feature. And yeah, as I mentioned, participatory process hosts are also using mainstream models like ChatGPT and Anthropic to brainstorm and develop participatory programs. And also, we at PeoplePowered are in the process of conceiving an AI tool that can help users to plan participatory processes. So, stay tuned for more on that maybe later this year or next year. Another example of AI before a participatory process is a dynamic FAQ and glossary. So, rather than pre-writing answers to the questions you expect people to have, you can aggregate common questions to your program's chatbot or other AI tool and publish answers to them. And you can see an example here from PopVox Foundation with their

[24:14](https://youtu.be/2sKTkE6R5OM?t=1454) StaffLink tool. AI during a participatory process. This is maybe the most exciting slide. Unfortunately, just because of the time limitations, I don't have time to go into all of this, but it's all in the guide. So, I definitely invite you to explore that to see everything that's possible. And I'm going to share a few screenshots of some examples. So, one of the most useful uses of AI is parsing, summarizing, and classifying participant contributions. And compared with traditional survey methods, generative AI's ability to parse large amounts of speech or text or other forms of engagement can allow hosts to ask more open-ended questions. And the AI feature allows administrators to handle large volumes of text, even if they receive 100,000 pages of transcripts, which I'm sure some of you have had to deal with that kind of quantity before. And actually, one note is that AI has been used for years to address this, even before the current gen AI boom. So, more mature participation AI features apply natural language processing to mine large volumes of text and classify discussion topics or estimate sentiment. And they can cluster wide-ranging conversations into a more finite set of topics for participants and process hosts. Another really common use that we're seeing is transcribing speech, as I mentioned before. And, yeah, so, it's not limited to working with text. And the more that platforms allow people to speak as they naturally would, the more accessible the processes become. And we actually have a case that we've written up

[26:19](https://youtu.be/2sKTkE6R5OM?t=1579) on our website from Germany's Siegen-Wittgenstein district, where urban planners piloted Deliberate, an AI tool, during their participatory social planning conferences. And this used a speech-to-text feature to improve documentation, note-taking, and it also saved time for facilitators to focus more fully on engaging with participants. Another common use is assisting participants through the process with bots. And I want to flag one risk here, is that different participants might have varying experiences with the same support bot, which can lead to unequal treatment. For example, a bot might provide two different people a different set of pros and cons about the same proposal. And because of privacy concerns, the process host might not know about these different experiences. So, this is something to keep in mind.

[27:20](https://youtu.be/2sKTkE6R5OM?t=1640) Those were just a few examples of AI during a process. There are also a few instances of AI being used after a participatory process. And one of the most interesting is sentence making, which helps participants and admins understand the results of a participatory process. So, you can see here that Assemble distills wide-ranging digital debates and represents them in visually creative formats, like mind maps, videos, even a comic strip summary. So, that's all AI generated, and it can make communicating about the results of the process a lot easier and engaging. Google's Jigsaw team also launched its own free product, SenseMaker, to provide this service. And then AI features have also been used to compose written summaries of the process and to collect and display analytics.

[28:21](https://youtu.be/2sKTkE6R5OM?t=1701) Okay. So, another check-in before we get to the final part of the presentation. At what stage or stages are you considering using AI to support a participatory process? Okay. We reached 50% participation in the poll. Okay. I think we can share the results now.

[29:21](https://youtu.be/2sKTkE6R5OM?t=1761) Okay. This is great. So, there's kind of a plurality for using AI before the process to help prep during the process and after the process. And then, yeah. Okay. So, yeah. I think, yeah. I think it'll be helpful for you guys to all check out the guide, because that provides some very specific examples as well as advice for how to use these features. Okay. So, now, before we go to the Q&A, I want to share This is a bit of a framework for you to help you evaluate digital platforms and AI features, as well as some resources that can help you with this. And it's important to take into account what we've spoken about so far, including the challenges and opportunities that AI presents, the principles that you want to approach it with, and then the different potential uses of AI features before, during, and after a process.

[30:29](https://youtu.be/2sKTkE6R5OM?t=1829) So, when selecting a tool, I think the most important thing to keep in mind, which I mentioned earlier, is think about the tool in service of your participatory process, rather than trying to make your participatory process fit into the shiny new tool. So, what you really need to do is assess and map your goals for the process, both in terms of what you want participants to get out of it, as well as policy outcomes that you want. Think about your available resources in terms of finances, human resources, technical know-how that you might need to set up and run the digital participation platform or AI feature. And then also think about your AI principles that you want to have in mind before you even think about the tools. And then based on that mapping, you can kind of identify which needs the digital participation tool and the AI feature can meet, both for the end user, which is the participants, and for the host, which is probably you as the program manager.

[31:44](https://youtu.be/2sKTkE6R5OM?t=1904) And then based on that needs assessment, you can identify the tool and the features that are most helpful to meet your needs. So, for example, do you need to make the process more efficient from the backend if you have limited human and financial resources to support administrative tasks and data processing? Or do you need to make the process more accessible for users who speak different languages, in which case you can use translation features, or for those who are less comfortable typing their answers, in which case you can use voice transcription, for example? And, okay, once you've mapped all of this out, goals, capacities, needs, from the perspective of the user and the host, I recommend that you use our ratings of digital participation tools to help you select what's right for your process. So we have, on our website, we have an analysis of 30 comprehensive tools that have been used to support digital participation globally. And this year we updated the ratings to include more information and insights about each tool.

[32:55](https://youtu.be/2sKTkE6R5OM?t=1975) We also made the air table where the ratings are a bit more interactive so that you can filter based on your needs. And then we also have some special mentions, which are digital tools that are doing a lot to leverage AI. So this can also be helpful when you're thinking about which tools specifically you want to use to harness AI for your process. And we also worked with Matt Stempeck, as well as our tech review committee, on the ratings this year, and we used research insights, expert review, and user and developer surveys. The ratings are based on six criteria. So you can also kind of think about your needs mapped against these criteria. So the criteria are platform cost, capacity requirements, meaning the technical know-how needed to set up and run a participatory platform.

[34:02](https://youtu.be/2sKTkE6R5OM?t=2042) The different features that are available through the tool. Accessibility in terms of different languages for people with disabilities, etc. Ethics and transparency. That includes the data policy, other security measures, is the platform open source, and then track record and reliability. So how long has the platform been available? How many countries is it used in? How many clients does it have? All these indicators of reliability. And yeah, so these are some questions to keep in mind when using the ratings. So is the tool cost effective, accessible, reliable, and appropriate for your context and program? What technical expertise is needed to use it effectively and efficiently? Does it offer support? Does it offer the features you require? And is it ethical in its data use, protection of personal information, and content moderation?

[35:09](https://youtu.be/2sKTkE6R5OM?t=2109) We also have these questions that are specific to AI features that you can ask when evaluating the different options. So think about which AI model the platform relies on and how that relates back to your AI principles. Does the data get shared with the AI provider, the participation platform, a third party that relates to privacy and security? At which stage of the process is the AI used? So thinking about at which stage in the process you would like to use it based on your needs. What does the platform use AI to accomplish? How will we know it's operating correctly and fairly? That relates to transparency. And then how does the AI perform with the languages spoken by the community? Because, you know, a lot of AIs are trained on like high resource languages. But if there are languages that are less commonly represented in tech, then the AI might not be as useful. So that's something actually quite important to keep in mind.

[36:21](https://youtu.be/2sKTkE6R5OM?t=2181) And then before I end, just to bring us to the Q&A, I do want to share a case study that we researched from Northern Ireland about using an AI-powered tool for engagement. So Northern Ireland is, of course, a post-conflict society. And the Voice Matters initiative used a digital participation tool provided by DemBrain that harnesses AI to capture speech. And basically, it allowed participants to speak more freely. It also freed up time for facilitators to pay more attention to the work of facilitation, like in-person facilitation. And like all of that kind of built greater transparency and trust in the process compared to previously when they weren't using this tool.

[37:27](https://youtu.be/2sKTkE6R5OM?t=2247) So this is actually a great example of technology being used in service of the participatory process and actually improving outcomes for participants and policy outcomes as well. So some of the lessons from this case are the importance of blending human facilitation, even when you're using digital tools, building trust through transparency and consent. Another part of it is the importance of clarity on data protection and privacy. Using accessible technology, a lot of users mentioned that it was really easy to use the tool. They just had to point their camera at a QR code, and then they were able to use DemBrain. And then finally, as I mentioned, choose a tool that aligns with your privacy and security principles. So I'm going to pause now so that we can do the Q&A. I'm really excited to hear what you guys are, what's on your mind.

* * *

Original page: https://innovate-us.org/different-ai-horses-for-different-courses-matching-tools-to-purpose · Licence: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/), as stated on innovate-us.org