This chapter analyses the opportunities for AI to address the challenges of citizen participation. It does so by proposing a typology of AI tools to improve and enhance citizen participation. The typology identifies nine types of applications and analyses their potential to support both governments and citizens in overcoming citizen participation challenges when designing, implementing, or participating in a citizen participation process. Building on the analysis of 50 case studies from 22 countries, this chapter also showcases emerging trends and discusses cutting-edge practices.
Artificial Intelligence and the Future of Citizen Participation
Typology of Applications, Opportunities and Challenges for Democratic Innovation
Abstract
Citizen participation opportunities are growing across OECD member countries and beyond, involving citizens at different stages of policymaking at the national and subnational levels (OECD, 2024[1]). However, the impact of these processes on policy and decision-making is often limited or difficult to assess. To strengthen citizens’ trust, providing additional opportunities for them to participate in is not sufficient. These opportunities should be meaningful and impactful, and they should hold governments accountable to their commitments.
The road to mainstreaming meaningful citizen participation is fraught with challenges. These challenges encompass a variety of dimensions: from shortcomings in the processes themselves (design, accessibility, scale) to their disconnection from their institutional environment (horizontal and vertical coordination, alignment between the “front-office” and “back-office” of participation), and finally to the deepening lack of mutual trust between governments and citizens. Understanding what lies beneath these challenges is key to identifying the areas where the use of AI can improve the outcomes and the impact of citizen participation processes.
Disconnection of citizen participation processes from policy impact and isolation from public debate
The impact of citizen participation is often hindered by its disconnection from policy and decision-making (OECD, 2025[2]; Loisel and Rio, 2024[3]). Moreover, the lack of feedback on the results of citizen participation processes and clear accountability mechanisms on their implementation might affect citizens’ trust in their impact (OECD, 2025[4]; Bertelsmann Stiftung, 2022[5]) These elements are the result of multiple forces, including a lack of political willingness, and the absence of shared understanding of citizen participation as a right and as a core government function. In addition, citizen participation processes are often implemented as ad-hoc initiatives instead of regular mechanisms embedded in institutional processes, which can result in unclear expectations and limited commitment in the implementation of their results in policy and decision-making (OECD, 2023[6]). Conversely, some processes are often implemented to comply with legal requirements, resulting in “tick-the-box” consultations (OECD, 2025[4]). According to the results of the 2023 OECD Trust Survey, the majority of citizens is rather sceptical about their voice being heard by their government when participating in a citizen participation process. On average, only slightly more than one in three citizens now think that their government would change a policy in response to a majority public opinion. Fewer than one in three citizens think that their government would adopt the opinion expressed in a public consultation (OECD, 2024[7]). Both figures have worsened on average compared to the 2021 OECD Trust Survey. Furthermore, the discussions held as part of citizen participation processes are often isolated from the broader public debate, which might hinder the legitimacy of policy outcomes that have an impact on the broader society.
Limited coordination, resources, and skills for participation
In addition to the disconnection from policy and decision making, the impact of citizen participation processes is hindered by poor coordination among institutions and across levels of governance, as well as by limited skills and dedicated resources. First, insufficient co-ordination in public institutions within and across levels of governance when organising, communicating, and implementing citizen participation processes can result in a multiplication of consultations on similar or overlapping policy issues. From the citizens’ perspective, such complexity can be difficult to navigate, triggering “consultation fatigue”, and ultimately leading to reduced participation and impact. Similarly, insufficient alignment between the “front-office” and “back-office” functions of government (see Box 2.1) can weaken the design and the impact of citizen participation processes (OECD, 2025[4]). Moreover, citizen participation requires dedicated skills, capacities, and resources in government (both human and financial) (OECD, 2023[6]; Bertelsmann Stiftung, 2022[5]). The absence, limited or discontinuous availability of these resources in governments might result in poorly designed and implemented citizen participation processes, ultimately undermining their impact in policymaking.
“Front office” functions of government include all public-facing activities such as public communications, responses to access to information requests, consultations, participatory processes and deliberative assemblies. These functions are generally more amenable to regular updates, tailoring to user needs and iterative development. With sufficient resources, the front office can be equipped to adopt new formats (in person/online), technologies (telephone/social media) and timings (synchronous/asynchronous) in response to citizen preferences.
“Back office” functions of government comprise internal processes such as public financial management and accounting, preparation of regulatory impact assessments, drafting of legislation and decrees, procurement, infrastructure planning, provision of in-service training, performance management and audit. These functions are generally ‘hard wired’ into public institutions and are often defined by specific laws or regulations, place high value on administrative certainty and are subject to strong scrutiny.
In reality, the distinction between front and back-office activities is often not always clear-cut for example, back-office activities might adjust to incorporate feedback collected through interactions with the public. However, it can be helpful to distinguish between front-office and back-office activities when describing citizen participation processes.
Governments should strive to align front and back-office on participation by adopting a citizen-centric perspective when designing, implementing, and reporting back on processes.
Source: (OECD, 2025[4]).
Design and accessibility challenges
Citizen participation processes are meant to expand and deepen the involvement of citizens in policy and decision-making. However, the design of these processes might not be adequate to provide meaningful and wide-reaching opportunities for all members of the public to get informed, understand the policy issues at stake, and participate (OECD, 2022[8]). The 2024 results of the OECD Trust Survey find that people with lower levels of education, younger people, and people identifying as part of discriminated groups (e.g. ethnic and linguistic minorities) systematically report lower levels of trust (OECD, 2024[7]). As citizen participation is meant to reinforce public trust, the design and implementation of processes should aim at involving as much as possible underheard groups all the while protecting the integrity of the processes from undue influence (OECD, 2025[2]). In addition, the impact of participatory and deliberative processes is often undermined by their limited scale. While some initiatives, such as representative deliberative processes, are designed to involve a limited number of participants, others that would benefit from broad inputs from a variety of stakeholders groups fail to mobilise a large audience, which can undermine the legitimacy of their results for policies and decisions that apply to society as a whole (Carson and Elstub, 2019[9]).
With the ongoing surge in the adoption of AI tools, scholars, experts, and practitioners in the fields of innovative citizen participation and civic tech have shown a lively interest in the opportunities created by these tools (Arana-Catania et al., 2021[10]; Asseh et al., 2025[11]; Berditchevskaia and Baeck, 2020[12]; Chemier and Hotsko, 2024[13]; García and Al., 2023[14]; Goldberg and al., 2024[15]; McKinney, 2024[16]; OECD, 2025[2]; Campagnucci et al., 2024[17]; Duberry, 2022[18]). Numerous experiments have focused on how AI tools can improve governments’ capacity to understand citizens’ needs and preferences through the analysis of qualitative inputs (Romberg and Escher, 2024[19]; Starke and Lünich, 2020[20]), improve the quality of online debates (Zhang et al., 2023[21]; Dos Santos, Saltz and Acosta, 2025[22]; Hadfi et al., 2021[23]), and enhance existing digital participation platforms (Arana-Catania et al., 2021[10]; Goldberg and al., 2024[15]; Small et al., 2023[24]). Overall, the literature acknowledges the potential of AI tools to address some of the challenges faced by citizen participation, by (1) improving the design and accessibility of citizen participation processes, (2) enabling the implementation of processes at scale while preserving their quality, (3) communicating better on citizen participation processes, bridging the gap between participants and the broader public debate (4) reducing costs associated with citizen participation processes, hence improving resource allocation. The literature also converge in affirming that some of the challenges of citizen participation, particularly those related to the disconnection of citizen participation processes from policy and decision-making, cannot be addressed by AI tools not by other digital and emerging technologies (OECD, 2025[2]). Finally, the literature discusses several common risks associated with the adoption of AI tools for citizen participation processes and stresses the importance of a careful and responsible uptake of these tools.
This report proposes a typology that goes deeper into these opportunities, highlighting how different tools can help address different challenges. However, its objective is not to simply encourage the adoption of AI tools in citizen participation process under the influence of technological hype (OECD, 2025[2]). It is also to help governments and practitioners to identify where, when, and how AI can empower and support them. It therefore rejects the techno-solutionist approach, which sees in technology a “quick fix” for complex societal problems (Morozov, 2013[25]; Duberry, 2022[18]), as well as the “solutionist trap”, which encourages the adoption of technological solutions without questioning whether they are relevant in context (Selbst et al., 2018[26]). On the contrary, the adoption of emerging technologies and AI tools in citizen participation should focus on creating opportunities for better interactions between government and citizens while promoting a healthy civic space in which civil freedoms are protected and the conditions for a thriving ecosystem of civil society organisation are met (OECD, 2022[27]) following the same logic of more mainstream civic technologies, such as citizen participation platforms. Moreover, this report envisions the role of AI as a support tool and not as a replacement for skilled practitioners and participants (OECD, 2025[28]) Finally, this report outlines the benefits of adopting AI tools as a means to expand citizens’ political agency in policymaking, and governments’ capacity to deliver effective policies in close collaboration with citizens.
AI tools for participation vary vastly in their forms and purposes. The following Typology aims at helping policymakers and practitioners navigate these tools and their uses in this context.
The Typology analyses how AI tools can support all government and citizen interactions as part of participation processes. It includes activities performed by governments and citizens to navigate and exchange information, as well as activities related to the design, implementation, and participation in citizen participation processes. Examples of citizens navigating government information are the access to government websites on policy issues and formats of participation. Examples of governments navigating citizen information are the social listening of online discourse to assess the main views of the public on policy priorities and options, as well as the analysis of citizen inputs, contributions, proposals, and projects submitted through formal channels of participation.
The typology builds on existing literature as well as on 50 real-world use cases from 22 countries collected through desk research. Based on this methodology, the findings in this chapter are not generalisable to the broader universe of AI uses in citizen participation. The findings do provide, however, observations rooted in real-world practice, as well as insights from the latest research. The Typology proposed is meant to navigate current AI opportunities for participation, while acknowledging that future and different practices and tools might emerge.
Structure of the Typology
The analysis of the literature and of use cases surfaced three criteria allowing the organisation of different existing and potential uses of AI tools to improve and enhance citizen participation. These three criteria are:
Types of Application: AI tools are first classified by their applications, meaning the tasks and activities carried as part of citizen participation processes that they can perform or support. For example, AI-powered tools capable of analysing vast amounts of textual inputs and highlighting trends or performing topic modelling are grouped under the application of “sense-making”. Conversely, AI-powered conversational interfaces (e.g. chatbots) that support citizens in navigating information are defined as “assistants”. The same application can intervene at different stages of citizen participation processes and potentially be used by both governments and citizens. Conversely, multiple applications may be needed to perform a specific activity or during a specific stage. This classification builds on the AI tasks identified in the OECD Framework for the Classification of AI systems (OECD, 2022[29]), and it adapts it to fit the context of citizen participation. The clustering of AI tools for participation by types of application presented in this report is a deliberate simplification, acknowledging that these categories are not mutually exclusive and that many tools encompass multiple functions concurrently. For instance, existing citizen participation platforms already integrate multiple tools that respond to a set of applications.
Front-office or Back-office: each application is analysed for the opportunities it generates in two dimensions of citizen participation, the government-citizen interface (“front-office”) and the internal work of public administrations (“back-office”) (see Box 2.1) (OECD, 2025[4]). For example, an AI-powered virtual assistant allowing participants to ask questions in natural language is used in the front-office dimension. Conversely a sense-making tool supporting civil servants in the analysis of citizens’ contributions to a public consultation addresses a back-office activity.
Potential to address citizen participation challenges: the various applications of AI tools included in this Typology tackle distinct participation challenges. For instance, a sense-making tool increases the productivity of civil servants when analysing citizen inputs, leading to a more efficient use of resources, while a virtual assistant improves the accessibility of a process by providing information in a user-friendly manner. While multiple applications can simultaneously address multiple citizen participation challenges, this Typology focuses on the main advantages of individual applications with respect to the challenges of citizen participation.
Figure 2.1 provides a visual representation of the Typology of AI tools for citizen participation. elaboration.
Table 2.1 provides more details on each type of application with respect to the front and back-office of participation.
Source: Author’s own elaboration.
Note: The Typology is meant to be updated with future applications depending on technological progress and uptake of new AI tools.
Source: Authors’ own elaboration
Types of applications
Information development
Designing and implementing citizen participation processes involves creating and disseminating information and communication materials. Quality communication is a prerequisite to organising a successful and impactful participatory process (OECD, 2022[8]). Citizens must be given the opportunity to get informed about the process, its recruitment criteria, the ways to participate, and the outcomes of their participation (OECD, 2021[30]). Effective information and communication efforts about these processes and their outcomes can contribute to incentivise citizens to participate and result in greater legitimacy, which, in turn, can improve public trust (Ardanaz, Otálvaro-Ramírez and Scartascini, 2023[31]). Beside the “how”, a citizen participation process should empower all interested publics regarding the “what”, by making the policy issues accessible and by allowing participants to form their own opinion.
AI tools, namely generative AI (GenAI) text, image, and video tools, can support governments in creating information and in adapting such content to various communication channels (OECD, 2021[30]; Lovari and De Rosa, 2025[32]). This can free up time for other tasks related to information dissemination, such as outreach to underrepresented categories, in-person communication opportunities and informative sessions. For example, the Government Communication Service of the UK is currently implementing a strategy for the safe and trustworthy use of generative AI in public communication (GCS, 2024[33]). Although the strategy is not specifically conceived for government communication on participation opportunities, its principles, such as “assess and tailor communications to make them increasingly accessible, helpful and relevant”, largely resonate with the objective of enhancing government outreach and improve accessibility of citizen participation processes. Information development tools are particularly relevant in contexts with limited resources dedicated to communication and participation, such as in local governments. For example, the Swindon Borough Council in the UK built the open-source AI powered tool “Simply Readable”, which leverages GenAI to adapt technical government documents into Easy Read format, improving the accessibility of their content for people with learning disabilities and visual impairment (AI.GOV.UK Knowledge Hub, 2025[34]).
Similar tools can also support citizens when participating, for example when drafting their contributions or in illustrating them through image generation. The use of AI tools when participating should be enabled and communicated as a way to provide a level playing field for the participation of all publics, for example by reducing self-censorship when drafting (Weeks, Halversen and Neubaum, 2024[35]), and not as a mean to replace citizens’ creativity and variety of opinions, and ultimately loss of autonomy and agency (Wihbey, 2024[36]).
Sense-making
Citizen participation processes aim primarily at creating channels beyond and between elections for citizens to express their views on policy issues and have a say in decision-making (OECD, 2022[8]; OECD, 2023[6]). The collection of citizen inputs, petitions, projects or proposals in participatory processes such as public consultations or participatory budgeting poses a two-fold challenge: for governments, analysing large amounts of qualitative contributions in a fair, exhaustive, and efficient manner, while for citizens and participants, engaging meaningfully with the contributions of their peers. For example, in 2019, the French national consultation (Grand Débat) gathered 1.9 million submissions through an online platform (French Government, 2019[37]).
The ability of AI systems in general, and of natural language processing (NLPs) models in particular, to make sense of large amounts of text can help governments analyse and summarise large volumes of citizen inputs, improving the efficiency of the process and the allocation of human resources. AI tools for sense-making allow for the automated annotation of inputs beyond the simple search for keywords, thus allowing to map recurrent topics, to cluster of opinions and ideas (Arana-Catania et al., 2021[10]) (Berditchevskaia and Baeck, 2020[12]), to detect outliers (Schneider and Sanders, 2023[38]). AI sense-making tools also allow for other types of analysis, such as the sentiment analysis of the perception of a given issue (Valle-Cruz et al., 2019[39]). Contributions to the French Grand Débat were analysed and clustered with the AI powered sense-making tool Arlequin AI (Asseh et al., 2025[11]).
Enhancing local public consultations in Cambridge with Insights by GoVocal
In 2024, the city of Cambridge in the UK opened a public consultation on the Design Code (a set of guidelines that shape the physical development of an area) of its northern neighbourhoods. Cambridge residents were invited to respond to an online survey and submitted geo-localised suggestions and comments on the local citizen participation platform built by civic tech firm GoVocal. Based on the results of this first phase, the city drafted five principles for improved neighbourhood design and shared the draft with residents to collect their feedback. The city used the AI-powered sense-making tool “Insights” embedded in the platform to analyse residents’ inputs and comments and to generate a comprehensive report. The City Council’s Consultation team estimated having saved about 50% of the time needed for the analysis of qualitative contributions.
Source: (Fillet, 2024[40]).
National Dialogues in Finland
In 2022, Finland started implementing National Dialogues, a participatory process aiming at creating a new model of social dialogue in cooperation with citizens, community and authorities, in order to build a share understanding of societal phenomena and issues that are important to the Finnish society. The National Dialogues rely on the Timeout method, developed by the Finnish innovation lab Sitra, which consists in multiple in-person facilitated sessions involving a group between 6 and 25 participants. As the National Dialogues are becoming a regular practice, the Finnish government is currently testing the use of AI sense-making tools to support the analysis of the qualitative inputs provided by participants and inform final reports and policy briefs.
When participating in a process, citizens might experience information overload (Perez, 2008[43]), feeling overwhelmed by the quantity of inputs provided by other participants. This might generate a multiplicity of similar proposals that then result in weaker impact or discourage participation. AI sense-making tools can help display content to better reflect the overall trends of other participants contributions’, flag the presence of similar inputs when drafting one’s own, or summarise the comments to a specific proposal (Arana-Catania et al., 2021[10]).
Improving public understanding of draft bills in Massachusetts - MAPLE
The State of Massachusetts in the US files more than 6,000 bills per year on average. To help citizens identifying, understanding, and commenting draft legislation on topics of interest, the non-profit organisation Massachusetts Platform for Legislative Engagement (MAPLE) deployed the MAPLE platform to help citizens of Massachusetts get informed, understand, and submit amendments and proposals for national legislation. Currently, the platform leverages AI-powered sense-making tools to provide users with bill summaries and to automatically assign topic tags to legislative proposals. The non-profit organisation is testing new features to compare and contrast similar legislation, support citizens when drafting testimonies (written feedback), analyse and summarize citizen testimonies on a same bill, and rank them to foster constructive dialogue and consensus.
Using NLP to help participants navigate contributions on Decide Madrid
In 2021, an experiment applied NLP techniques to improve the participants’ experience on the participatory platform of the city of Madrid in Spain, Decide Madrid, based on the open-source software Consul. The experiment sought to reduce information overload by (1) suggesting other participants’ proposals in line with one’s interest, (2) grouping citizens by interests, and (3) summarizing the comment sections of existing proposals. Results showed reduced information overload and enhanced participants’ interactions. It also highlighted the need for further fine-tuning of the AI-enhanced version of the platform, particularly for the summarization of comments. A new project to equip Consul with an AI-powered Civic Assistant was launched in early 2025 and is currently at the research stage
One of the challenges faced by citizen participation processes is their isolation from broader society (OECD, 2025[47]). AI tools can play a significant part in making the results of participatory and deliberative processes more accessible to the broader public, including journalists, civil society, elected representatives and political parties. Such enhanced accessibility could result in increased awareness and resonance of the policy issue covered by the public debate, potentially leading to greater policy impact. For example, in 2024 the Economic, Social and Environmental Committee (CESE) of France adopted the tool Panoramic.AI developed by Make.org to “open up” the proceedings and the results of the Citizen Convention on End of Life (CESE, 2024[48]; Combaz et al., 2024[45]). The platform allows users to formulate queries in natural language and references videos and written records from the Convention.
Translation
Language can constitute a significant barrier to the participation of linguistic minorities in each country or territory. Moreover, multilingual, supranational, and international participatory processes are often confronted with the complexity of producing information content and enabling the participation of citizens in multiple languages. AI-powered automatic translation systems are already opening new opportunities for multilingual participation. For example, the European Commission organised the Conference on the Future of Europe allowing participants to submit their inputs in all 24 official languages of the European Union thanks to the AI-powered eTranslation tool (European Commission, 2021[49]; Fabbrini, 2021[50]).
For citizens, high-quality automated translation results in being able to interact with other participants writing in their own language, creating a level-playing field for meaningful dialogue. Translation is one of the measures that governments can adopt to ensure the accessibility of citizen participation processes to underrepresented and marginalised groups, such as immigrants and linguistic minorities. The Civic Engagement Commission of the City of New York (USA) adopted the participatory platform Participate New York based on the open-source software Decidim, which allows New Yorkers to access information and participate in all languages thanks to the Google AI-powered automatic translation tool (Civic Engagement Commission, 2025[51]). For governments, automated translation tools enable multilingual processes without increasing costs.
Translation into sign language can significantly improve the experience of citizens with hearing impairment. Governments are already adopting AI-powered video generation tools to translate voice and video messages into sing language through synthetic signers in public spaces or in the delivery of public services. For example, federal participatory platform of Brazil, Brasil Participativo, embedded the open-source AI-powered avatar VLibra that translates website contents into sign language (Brasil Participativo, 2025[52]).
Transcription
AI tools can also contribute to lowering other barriers to access to information and to participation. For instance, AI-powered tools that transform text inputs into speech and spoken inputs into text have multiple potential uses for citizen participation. For participants with visual impairment and other reading impediments, text-to-speech tools enable them to get informed about the policy issues at stake and about the rules of the processes they participate in.
Governments and practitioners can use speech-to-text tools to collect participants’ spoken contributions, during in-person events and better include citizens with insufficient digital skills or without access to digital devices. For example, in 2019, the city of Lugano (Switzerland) used “Project Debater - Speech by Crowd” developed by IBM to collect and transcribe oral contributions to a local consultation on self-driving vehicles. The tool allowed to collect and analyse around 2500 statements (Lugano Living Lab, 2019[53]). During deliberative processes, speech-to-text tools can help facilitators keep track of the arguments raised during the sessions and better structure report drafting. In Belgium, facilitators of the Citizen Panel on Artificial Intelligence held in 2023 recorded the discussions among participants and transcribed them with an AI-powered speech-to-text tool, which enhanced the accuracy of documentation while significantly reducing the time spent for transcribing and summarising (Belgian Presidency, 2024[54]). By using transcription tools, governments and practitioners can both expand the reach of participatory processes, as they enable new channels for citizens’ inputs, and better allocate resources when conducting live activities, as they reduce notetaking of discussion points.
Virtual assistance
Citizen participation processes can be demanding of citizens, leading to self-exclusion among those unfamiliar with government information or who find it too complex to navigate. For instance, understanding and formulating an opinion about the policy issues at stake, keeping track of the calendar of specific participatory opportunities, or using digital platforms can be challenging in everyday life of ordinary citizens (OECD, 2025[4]). Virtual assistants such as chatbots, defined as conversational interfaces supporting users in navigating information and/or providing clear paths of action, can support citizens in making sense of complex participatory processes by answering their queries in plain language. They can also send personalised notifications and reminders based on their interests (Androutsopoulou et al., 2019[55]) (Van Noordt and and Misuraca, 2019[56]) (Cortés-Cediel, 2023[57]). Chatbots have existed long before the advent of Generative AI, but their capabilities were limited to responding to simple rule-based queries. AI-powered chatbots now perform complex tasks and can build new communication channels between governments and citizens (Androutsopoulou et al., 2019[55]) The city of Madrid in Spain experimented with a chatbot virtual assistant called Clara on its participatory platform Decide Madrid. In some cases, chatbot interfaces can also become a channel for easier participation, as with the open government chatbot, Chatico of the city of Bogotá in Colombia. By adopting virtual assistance tools that support citizens in navigating and remaining informed about participation opportunities and related procedures, governments can improve resource allocation, potentially lowering the number of information requests requiring staff attention.
Chatico Bogotá
The city of Bogotá in Colombia developed the AI-powered chatbot “Chatico”, a virtual assistant that helps residents engage with the participatory platform Gobierno Abierto Bogotá. Residents can also use it to report on matters of corruption and urban management. The city made the chatbot accessible via WhatsApp which allows users to receive “Chatico” updates directly on their devices.
Clara
The participatory platform Decide Madrid deployed the virtual assistant Clara, an AI-powered Chatbot that support citizens in understanding how to participate in online and offline participatory processes launched by the city of Madrid.
Virtual assistants can also support civil servants in preparing contents and digital spaces for citizen participation. For example, German startup demokratie.today created an AI-powered virtual assistant allowing administrator of digital participation platforms based on the open-source software Consul to easily set up digital online participatory processes (Demokratie Today, 2025[60]).
Moderation
The experience of digital citizen participation can be affected by harmful content or spam. All online environments enabling interactions among users, defined by some scholars “digital public squares” (Goldberg and al., 2024[15]), such as social media platforms, have adopted content moderation policies defining their code of conduct and redress mechanisms. Online content moderation can be defined as screening, evaluating, categorising, approving or removing/hiding online material according to relevant communications and publishing policies. It seeks to support and enforce positive communications behaviour online, and to minimise aggression and anti-social behaviour, and ultimately reinforce trust and safety (Flew, Martin and Suzor, 2019[61]). Although all forms of content moderation pose significant challenges and controversies (De Gregorio, 2020[62]), for example related to protecting freedom of speech or avoiding the silencing of specific categories or opinions (Dias Oliva, Antonialli and Gomes, 2021[63]), they represent an “essential, constitutional, definitional aspect of what platforms do” (Gillespie, 2018[64]). Current online moderation on social media platforms such as Facebook, Instagram, and YouTube, is often the result of automated and non-automated processes (Roberts, 2017[65]).
In addition, government websites, including participatory platforms, are often subjected to Black Hat SEO attacks, a spam technique that targets high ranked websites on research engines to artificially inflate the web performance of other websites (Yang et al., 2021[66]). These attacks flood institutional websites with spam content, polluting the quality of digital spaces. The presence of spam material, including harmless one, can negatively impact citizen engagement (Dasgupta et al., 2012[67]).
AI spam detection and content moderation tools can support civil servants acting as community managers in keeping the online spaces dedicated to citizen participation safe and constructive environments for discussion and collective decisions. In the same spirit, newspapers such as the New York Times, the Financial Times and Le Monde have opted for the open-source AI tool Perspective API developed by Google’s technology incubator Jigsaw to flag and remove toxic content from their websites’ comments sections and foster constructive dialogue among readers (Dos Santos, Saltz and Acosta, 2025[22]).
TrollWall – Office of the President of the Slovak Republic
The office of former President of the Slovak Republic Zuzana Čaputová used the AI moderation tool TrollWall, a proprietary solution developed by the eponymous Norway-based startup, to shield the former president’s social media accounts from hate speech and “trolls”.
Facilitation
In participatory and deliberative processes, facilitation occupies a prominent role (Dillard, 2013[69]; Moore, 2012[70]). Facilitation is the practice of enabling groups to work together cooperatively and effectively, voice individual opinions in constructive ways, and work together towards consensus, mutual understanding, and shared decision-making (OECD, 2020[71]; White, Hunter and Greaves, 2022[72]), with a strong emphasis on ensuring that all participants are meaningfully involved in the process (Fung, 2005[73]; Smith, 2009[74]). The role of a facilitator is not to provide content or direct outcomes, but to guide the process in a way that allows participants to form their own opinion and contribute their knowledge, perspectives, and ideas to the discussion (Romberts and Escobar, 2015[75]). Facilitation consists in a range of activities and techniques, from time-keeping and fair distribution of speaking time, to understanding and making sense of the discussion, to supporting participants in finding common ground and reach consensus (Stromer-Galley, 2007[76]).
Facilitating live participatory or deliberative processes, whether they take place in person or online, is constrained by issues of scale. The literature suggests that the ideal size for a facilitated group discussion varies between 6 and 12 participants depending on the objective, the number of questions, the time allocated, and the format of the exercise (Tang and Davis, 1995[77]; Byers, Zeller and Byers, 2021[78]; Cortini, Galanti and Fantinelli, 2019[79]). In deliberative processes it is generally recommended to split the group of participants in sub-groups of 6 to 8 people (White, Hunter and Greaves, 2022[80]). According to Landemore, digital technologies and AI tools in particular could help resolve the trade-off between democratic deliberation and mass participation (Landemore, 2022[81]).
AI tools can support participatory and deliberative processes by helping structure discussions, moderate exchanges, and clarify complex debates. For example, researchers at Google DeepMind created the Habermas Machine, an LLM-powered facilitator, which improved clarity, informativeness, fairness, and consensus-building in a UK virtual assembly of 5,734 participants (Tessler and al., 2024[82]). Several controlled experiments have shown the benefits of AI facilitation tools to make discussions more constructive , include all participants, and build consensus on divisive policy issues (Argyle et al., 2023[83]; Wyss and Beste, 2017[84]; Kim et al., 2021[85]; Hadfi et al., 2021[86]; Ito, Hadfi and Suzuki, 2021[87]; Zhang et al., 2023[21]). Box 2.6 gives more examples of AI facilitation tools for participatory and deliberative processes.
LXS 400: Chile Delibera
In 2021, the Chilean Senate partnered with the Stanford Deliberative Democracy Center and the NGO Tribu Foundation to organise a representative deliberative process involving 514 Chileans for three days to discuss policy priorities selected in a previous national consultation process, i.e. pensions and healthcare. Given the context of the pandemic, the Chilean Senate opted for an online process. The deliberative sessions took place on the Stanford Online Deliberation Platform, which uses the AI facilitator Alice to equally distribute speaking time and encourage all participants to take part in the discussion.
A safer and fairer Derry/Londonderry: Voice Matters project
In April 2025, the NGO Co-operation Ireland (Northern Ireland, UK) hosted the third edition of the Voice Matters project in Derry/Londonderry, a three-day community dialogue to reflect with residents on how to make the town safer and fairer to all. For the 2025 edition, facilitators used the open-source AI tool Echo developed by the startup Dembrane. Echo transcribes live conversations to surface discussion topics and opinions and prepare draft reports.
EU Committee of the Regions – AI enhanced workshop on Citizen Engagement
In March 2024, as part of the Opening Conference of its 2025 Young Elected Politicians Programme, the Committee of the Regions of the European Union organised a workshop on citizen engagement between election cycles. To support the work of facilitators, the organisers tested the AI-powered tool DeliberAIde, developed by the eponymous German startup, specifically for citizen participation processes. DeliberAIde performs a variety of tasks including assisted participatory process design, anonymised transcription of discussions, summarisation, topic modelling, visualisation, and report drafting.
Simulation
Public authorities at different levels of government engage the public in re-imagining and re-building physical spaces, e.g. through participatory processes that select various options for repurposing existing infrastructure (OECD, 2023[93]). However, it is difficult to achieve a shared understanding of the potential outcomes of the policy interventions available without a way to describe or visualise them in a shared space. AI tools, especially image generation ones, unlock new opportunities to simulate and visualise future and alternative scenarios (Berigüete, 2024[94]; Otherngrafen, Sievers and Reinecke, 2024[95]), by improving the design and accessibility of citizen participation processes (Guridi et al., 2024[96]). More specifically, generative AI can significantly reduce the costs of creating visual simulations by automating the design process and generating realistic outputs (Preussner et al., 2025[97]). By using simulation tools, governments could design more engaging and meaningful participatory processes. For example, in urban planning, local governments, urban planners, and citizens can prompt in natural language, visualise, and provide feedback on different options for a given space (Williams et al., 2024[98]). In 2023, the city of Helsinki in Finland used the AI simulator UrbanistAI to engage with citizens and local entrepreneurs on repurposing three streets in the centre of the capital, that would remain closed to car traffic in the summer months (UrbanistAI, 2023[99]; City of Helsinki, 2025[100]).
AI simulation tools can also prove useful when discussing long-term policy issues requiring foresight and scenario-scanning (Fitkov-Norris and Kocheva, 2025[101]; Polchar and Karle, 2023[102]). Such innovation is particularly suitable to support practitioners and participants while discussing long-term policy issues (Weyl, Tang and Community, 2024[103]). For example, the United Nations Development Program’s (UNDP) Panama Acceleration Lab used UrbanistAI to involve the students at the University of Panama in a scenario-planning exercise on the consequences of raising temperatures and sea levels (Chemier and Hotsko, 2024[13]), whilst the Civic Innovation Lab of the NGO BetaNYC developed the advocacy tool FloodGen, allowing New Yorkers to visualise the consequences of potential flood scenarios (Kittredge, 2024[104]). Beyond image generation, AI simulation tools make it easier to create immersive 3D models and interactive virtual simulations (digital twins) (Quan, 2022[105]; Xu et al., 2024[106]; OECD, 2025[107]).
Architecture of democratic digital spaces
Artificial intelligence can be used to shape digital spaces to promote constructive dialogue, foster democratic values, and harness collective intelligence (Weyl, Tang and Plurality, 2023[108]; Ovadya, 2023[109]; Duberry, 2022[18]; Verhulst, 2018[110]). Scholars have defined this application of AI as “collective dialogue systems” (Goldberg and al., 2024[15]). Ovadya puts them into a broader category he calls “bridging systems” (Ovadya, 2023[111]). He contrasts them with existing digital spaces whose architecture may foster mis- and disinformation, as well as polarisation (OECD, 2024[112]). For instance, open-source platforms like Polis and Talk to the City use machine learning techniques to map participants’ positioning in online conversations with the aim of encouraging mutual understanding and nuanced perspectives on a given conversation topic (Small et al., 2023[24]). AI-powered platforms designed to promote constructive dialogue are an emerging trend in civic technologies development (People Powered, 2025[113]). These platforms also have a role to play in bridging small-scale citizen participation processes with the broader public debate. For example, the Austrian Citizens’ Climate Assembly of 2022 used Polis to involve larger swaths of the public in providing feedback to the first draft recommendations. About 6,000 people participated in the online discussion (Paice and Rausch, 2022[114]).
What do you think, Finland?
In 2024, the Finnish Innovation Lab Sitra organised the public consultation “What do you think, Finland”? using the open-source AI-powered platform Polis. Over 18,000 participants expressed their agreement, disagreement or neutrality on “seed” statements on a variety of policy issues and submitted their own contributions. The software, significantly improved in its UX design by DigiFinland and fine-tuned to perform adequately in the Finnish language, made it possible to map opinions and sentiments.
“What do you think, Finland?” was Sitra’s first experimentation of Polis at the national level. Since 2023, the Innovation Lab has been experimenting with Polis at the local level, partnering with the wellbeing services of the South Karelia, North Karelia, Pirkanmaa, Central Finland, and Central Ostrobothnia counties for a total of more than 12,000 participants. These pilots are part of Sitra’s broader strategy to embed citizen participation in policymaking.
Sense-making, virtual assistance, automated translation tools are becoming widespread
The analysis of use cases carried out for this report shows that AI sense-making, virtual assistance, and automated tools are the most widespread applications among the ones described in the Typology (see Figure 2.2). These results resonate with an in-depth analysis of 31 participation platforms conducted by Civic AI (Zisengwe and Stempeck, 2025[118]), as well as with the trends identified by the Scope research project of the University of Münster (Campagnucci et al., 2024[17]).
Note: N=50.
Source: Author’s own elaboration.
In the case of AI translation tools, it is plausible to assume that their use for citizen participation processes is rather a natural extension of their widespread use in digital spaces. The adoption of AI virtual assistants underscores governments’ efforts in providing citizens with more accessible information in a more efficient manner. Sense-making tools, especially when used to analyse citizens’ inputs, contributions and proposals, seem to respond to the need of governments to improve their understanding of citizens’ voice on given policy issues while enhancing the efficiency of the process. For example, the AI incubator of the UK Government is currently developing and testing the sense-making tool Consult to improve the national public consultation process (see Box 2.8).
The UK government runs on average 600 public consultations per year. Analysing participants’ inputs requires significant human and financial resources (on average, a consultation that receives 30,000 responses requires a team of 25 analysts for a period of three months). The UK government’s AI incubator (iAI) sought to improve this process by creating the AI-powered open-source tool Consult. Consult uses AI and data science techniques to automatically extract patterns and themes from the responses and to turn the results into dashboards. In 2025, iAI tested Consult to analyse the responses to the Scottish’s government consultation on non-surgical cosmetic procedures. Those tests gave positive results on the tool’s ability to identify themes, allowing analysts to speed up the review process. Consult also helped reduce reviewer bias. However, the tool struggled to identify missing themes and analysts claimed more agency over specific steps of the process. If deployed in all consultations, Consult is estimated to save £20m and 75,000 hours per year.
Using AI to enhance and scale deliberation is a priority
The literature indicates strong interest in developing and adopting AI tools to enhance representative deliberative processes (McKinney, 2024[16]; Weyl, Tang and Plurality, 2023[108]; Sarabi, 2025[121]). Representative deliberative processes are citizen participation processes that bring together a group of citizens selected through sortition to represent society at large to get informed, discuss in-depth, and deliberate on a given policy issue (OECD, 2020[71])1. The impact and the legitimacy of these processes are often challenged by their limited scale, their isolation from the broader public debate, and the resources needed to implement them (OECD, 2025[2]; Landemore, 2022[81]; Ehsassi, 2024[122]; Niemeyer, 2014[123]). There is no unidimensional definition of “scaling” deliberation and other democratic innovations to enhance their impact. A study performed a meta-analysis of 300 democratic innovation projects to consolidate a theory of scaling including four dimensions: embedding democratic innovations in institutions and decision-making (scaling high), rooting participation in culture (scaling deep), engaging more citizens (scaling out), improving the quality of democratic innovations (scaling in) (ScaleDem, 2025[124]). The think tank DemocracyNext adopted a similar framework to focus exclusively on representative deliberative processes and identify the areas of opportunity for AI tools (Chwalisz and McKinney, 2025[125]), while McKinney, Schäfer and al., and Goldberg and al. mapped the potential of AI applications throughout the journey of a representative deliberative process (McKinney, 2024[16]; Goldberg and al., 2024[15]; Schäfer, Scheunemann and Salecker, 2025[126]). Figure 2.3 builds on these frameworks to provide examples of relevant AI applications in deliberative processes, while Box 2.9 details an example of AI-enhanced assembly conducted with students in Deschutes County, Oregon (US).
Different AI tools can support organisers, facilitators, and communicators throughout the implementation of a representative deliberative process.
Note: The figure provides an overview of opportunities for AI tools to support the preparation, implementation, and communication of a representative deliberative process. It does not suggest adopting every tool listed for every activity.
Source: Author’s own elaboration based on McKinney, Goldberg and al., and Schäfer and al., People Powered (McKinney, 2024[16]; Goldberg and al., 2024[15]; People Powered, 2025[127]).
Deschutes Civic Assembly on Youth Homelessness
In late 2024, civil society organisation Healthy Democracy partnered with the Central Oregon Civic Action Project, the think tank DemocracyNext, and the MIT Center for Constructive Communication to launch the Deschutes Civic Assembly on Youth Homelessness, a pressing issue in Deschutes County, Oregon (US). Healthy Democracy sent an invitation letter to 12,500 residents and selected a panel of 30 delegates assessed as representing a culturally and demographically representative panel for Deschutes County. The Assembly convened for two weekends of facilitated learning and deliberation to collectively respond to the question “What should our priorities be for building community solutions to prevent and end youth homelessness?” and provided 23 recommendations with more than 77% approval rate ranging from governance structures, funding suggestions, and awareness campaigns. The Assembly benefitted from the AI platform developed by the non-profit Cortico AI, which combines multiple applications (transcription, sensemaking, and information development) to support facilitators during and after the deliberative sessions.
Beyond application-based AI tools: AI agents and AI-human policymaking
The typology classifies tools by individual applications (tasks and activities) to provide an overview of existing tools and support governments identifying relevant opportunities when organising and implementing citizen participation processes. However, evidence from use cases shows that such tools can be combined and organised into seamless workflows tailored to support the specific needs of participatory and deliberative processes. For example, a consortium led by Consul Democracy Foundation is currently developing a suite of AI applications to enhance the open-source participation platform Consul, used by more than 250 cities and organisations around the world (Consul Democracy Foundation, 2025[130]).
Although most of today’s AI tools can be considered narrow and designed to perform a specific task, early forms of “agentic” AI are emerging (OECD, 2025[28]). This refers to systems that can manage end-to-end processes, learn, self-optimise, and collaborate with humans and other agents with minimal human input (Ilves et al., 2025[131]; Government Technology Agency Singapore, 2025[132]). Simple examples include agents capable of autonomously browsing the web, interpreting the results of research, and offer selected alternatives to users. Due to their more autonomous and complex nature, agentic AI introduces new risks and challenges that should be weighed carefully (Iason et al., 2025[133]). The current state-of-play of existing agentic applications for participation remains very limited. However, future and more advanced AI agents designed specifically to support citizen participation could help civil servants through all phases of participatory and deliberative processes, enabling a better allocation of resources, and an expansion of the capacity of governments to engage with citizens. Moreover, future AI agents owned and configurated by citizens could represent them in participatory processes by voicing their pre-selected opinions and preferences (Albrecht, 2025[134]).
Considering the fast-paced evolution of AI technologies, it is relevant to start a reflection on the impact of the use of AI in policy and decision-making through the lens of participation. Governments across OECD member countries and beyond are already using AI to make predictions and forecast future scenarios, to formulate and evaluate policy alternatives, and to optimise resource allocation (OECD, 2025[28]). In a world of AI-enhanced policymaking, governments should seize the opportunities created by AI technologies to envision the role citizens can play as key contributors to effective and democratic decisions and policies in new ways and at greater scale. For example, the Governance Lab (US) partnered with Citizens Foundation (Iceland) to develop Smarter Crowdsourcing, a human-AI collective intelligence methodology aiming at harnessing expertise in society to respond to societal and policy issues. The methodology uses the toolkit Policy Synth, comprising several AI applications (Bjarnason, Gambrell and Lanthier-Welch, 2024[135]).
Beyond participation: AI opportunities to foster government openness, responsiveness, and accountability
Citizen participation is one of the core principles of open government, together with transparency, integrity, and accountability. Open government is defined as a culture of governance that embraces and promotes these principles (OECD, 2017[136]). Access to quality government information and data, effective government communication, and a protected civic space are essential prerequisites for meaningful and impactful citizen participation (OECD, 2023[6]). AI presents relevant opportunities in all these areas. Citizens could benefit from AI interfaces to navigate the complexity of government information, while governments would expand their capacity to understand public debate, and listen to citizens’ voices, including beyond and between dedicated citizen participation processes.
Helping citizens and civil society navigate the complexity of government information
AI tools can help citizens navigate the complexity of government information by making it more accessible, intelligible, and actionable (Berryhill et al., 2019[137]; OECD, 2025[28]):
Open government data: AI-powered sense-making and virtual assistance tools can analyse and visualise open government data, improving their re-usability (Proscovia, 2025[138]). For example, in Washington D.C. (US), the District of Columbia’s Enterprise Data team is building DC Compass, an AI sense-making tools that allows users to create interactive maps based on open government data.
Government websites: Virtual assistants and conversational agents embedded in government websites can guide citizens towards relevant information and services (Cardoso, 2023[139]). For example, the Government Digital Service of the UK is currently developing GOV.UK Chat, a chatbot designed to provide information on business and trade. Findings from the experimental phase show that 70% of users found the responses useful (Gregory et al., 2024[140]). The Italian National Social Security Institute (INPS) deployed Arianna, a virtual assistant allowing users to understand and compare pension and social security schemes (INPS, 2025[141]).
Enhancing government transparency and accountability
Governments can use AI tools to support the enforcement of transparency regulations and provide citizens with accessible information on the implementation of policy. Likewise, citizens and civil society organisations can leverage AI tools to expand their oversight and monitoring capacities:
Reactive and proactive disclosure of government documents: AI tools can support analysts to process and respond to access to information requests (reactive disclosure), and to reduce the time and cost of reviewing, classifying, and disclosing government documentation (proactive disclosure). In 2023, the US government adopted the AI-powered sense-making and information development tool MITRE to support analysts in processing and responding to FOIA (Freedom of Information Act) requests (Gast Romps, 2023[142]).
Legislative archives: AI sense-making tools can support citizens and stakeholders in understanding the background of existing regulation by improving the accessibility of parliamentary archives. The European Parliament added a GenAI powered interface to its archive’s website, including documents from 1952 to 2009, allowing users to explore them online through natural language requests (European Parliament, 2025[143]).
Monitoring of parliamentary activities and political debates: AI-powered transcription, information development, and sense-making tools can be powerful allies of citizens, civil society, and journalists to stay updated with legislative and political debates. The municipalities of Argos, Xylokastro, and Chania in Greece adopted the tool OpenCouncils developed by the non-profit civic tech organisation SchemaLabs to record, transcribe, analyse, and draft reports of their City Council meetings (OpenCouncil, 2025[144]). Similarly, the Brazilian Chamber of Deputies adopted Ulysses, a Suite of AI tools improving the tracing and the transparency of parliamentary works, including a sense-making tool to analyse citizens’ feedback (Bussola Tech and Luís Kimaid, 2024[145]).
Enabling social listening and improving government communication
Government communication is defined as the function through which government delivers information, listens, and responds to citizens in the service of the common good (OECD, 2021[30]). Effective government communication is a powerful driver of policy acceptability (OECD, 2025[146]) and public trust (OECD, 2024[7]). AI sense-making and information tools can enhance the capacity of public officials to listen to citizens and tailor accessible and effective communication efforts.
Social listening: AI-powered sense-making tools can help public communicators monitor and analyse online debate, audience perceptions, and attitudes before policy issues (Kumi et al., 2024[147]; Arunachalam and Sarkar, 2013[148]). In 2019, UN Global Pulse used transcription and sense-making AI tools to analyse radio conversations on public policy and government action in Uganda (Duberry, 2022[18]; Rosenthal, 2019[149]).
Tailored effective communication: AI tools can enhance the efficiency and effectiveness of public communicators by assisting in brainstorming, automating routine tasks such as media planning, and ensuring consistency in content across communication channels (Berryhill et al., 2019[137]). These capabilities enable communicators to better engage with their audiences and tailor messages to diverse groups. For example, the Government Communication Service of the UK developed Government Communication Service Assist, a suite of AI-powered tools based on Microsoft Co-Pilot that support all activities of government communicators. Assist is the result of two years of successful experimentation and testing (Wade, Chahal and Melton, 2025[150]).
Enhancing government responsiveness through personalised and interactive public services
Public services represent a fundamental component of the citizen-government interface. The quality of public services has a profound impact on people’s lives and is often pivotal in ensuring citizens have access to opportunities and realise their full potential (OECD, 2024[151]). The positive perception of public services is also associated with higher levels of trust in government (OECD, 2024[7]). Governments are leveraging AI to improve and personalise the delivery of public services. For example, Italy started using AI to sort and classify web requests to its National Institute of Social Security (INPS), which increased dramatically during the Covid-19 pandemic. With a focus on precision, more than two-thirds of emails are now automatically dispatched, with 80% accuracy (OECD, 2024[151]).
Protecting the civic space and supporting civil society organisations
Civic space is defined as “the set of legal, policy, institutional, and practical conditions necessary for non-governmental actors to access information, express themselves, associate, organise and participate in public life” (OECD, 2022[27]). A healthy and protected civic space is today widely recognised as an essential precondition for democracy and participation (OECD, 2023[6]). The research and policy debate at the intersection of AI and civic space focuses primarily on the threats that AI tools pose to civil freedoms, namely automated surveillance, mis- and disinformation, privacy infringements, and censorship (OECD, 2024[152]). Nevertheless, citizens, civil society, and human rights defenders can use AI tools to support their advocacy and data collection activities.
Smart crowdsourcing: AI tools can help citizens, journalists, and civil society organisations to analyse crowdsourced and geospatial data on electoral incidents, attacks and violence against human rights defenders and peaceful assemblies (Nesta, 2021[153]). The Brazilian Association of Investigative Journalism (Abraji) and Mexican data journalism nonprofit Data Crítica developed the open-source AI sense-making tool "Attack Detector," which uses NLP to detect and monitor hate speech against journalists and political figures (Linares, 2023[154]). Similarly, in 2018 the NGO Amnesty International collaborated with Element AI to develop a monitoring tool providing updates on death penalty cases (Wildi, Braizat and Evrard, 2024[155]). Beyond monitoring existing data, AI tools can be used to engage with citizens in complex contexts to surface their priorities and needs. For example, between October 2020 and January 2021, the UN Department of Political and Peacebuilding Affairs (DPPA) used the AI-powered platform Remesh to conduct five live polls to surface Libyans’ views on the country’s civil war and future perspectives (Masood Alavi, 2022[156]).
Advocacy and campaigning: AI sense-making and information development tools can support small organisations, individual citizens, and non-profit organisations to monitor the political debate and build advocacy campaigns and communication strategies with limited resources (Civic Tech Field Guide, 2023[157]).
CSO and community-lead participation processes: Civil society organisations and self-organised communities can benefit from the applications analysed in the Typology to conduct their own participation processes and engage with their communities. For example, the CSO Silent Cry from Michigan (US), which provides mental health support to inmates, used the AI-powered platform Talk to the City to analyse the insights of a collection of video interviews with formerly incarcerated individuals to surface their challenges when reintegrating society (Talk to the City, 2023[158]).
Awareness raising about citizens’ rights: AI-powered virtual assistants can play a pivotal role in informing people about their legal rights in a context-sensitive and accessible manner. For example, the Canadian non-profit organization People’s Law School launched the virtual assistant Beagle+, powered by ChatGPT4 by OpenAI, to help British Columbia’s citizens obtain information on their rights and legal obligations (People's Law School, 2025[159]).
AI for electoral management
The OECD focuses on the involvement of citizens beyond electoral representation. Nevertheless, governments could get inspiration from the opportunities unlocked by AI tools in the context of elections. A report by International IDEA (Institute for Democracy and Electoral Assistance) analysed how AI tools can assist Electoral Management Bodies (EMBs) throughout the electoral cycle (Juneja, 2024[160]). Before elections, AI tools can be adopted to perform voter list management, voter registration, resource allocation planning, election costs forecasting, or targeted advertising. During elections, AI tools can contribute to campaign and media monitoring, biometrics, as well as voter verification, polling place monitoring, and ballot counting (Dwyer, 2024[161]). Finally, artificial intelligence can support post-electoral audits. The use of AI to support electoral management presents significant ethical, human rights, and practical concerns, ranging from technically difficult implementation to serious issues with electoral integrity and public trust.
References
[126] (Hrsg.), J. (ed.) (2025), KI-gestützte Bürgerräte - Eine Chance für mehr Inklusivität und Gleichberechtigung in der Demokratie, https://doi.org/10.14512/9783987264511.
[119] AI.GOV.UK (2025), Consult, https://ai.gov.uk/projects/consult/.
[34] AI.GOV.UK Knowledge Hub (2025), Simply Readable, https://ai.gov.uk/knowledge-hub/use-cases/simply-readable/.
[134] Albrecht, E. (2025), Political Automation, Oxford University Press, https://global.oup.com/academic/product/political-automation-9780197696958?cc=fr&lang=en&.
[55] Androutsopoulou, A. et al. (2019), “Transforming the communication between citizens and government through AI-guided chatbots.”, Government information quarterly, Vol. 36/2, pp. 358-367, https://doi.org/10.1016/j.giq.2018.10.001.
[10] Arana-Catania, M. et al. (2021), “Citizen participation and machine learning for a better democracy”, Digital Government: Research and Practice, Vol. 2/3, pp. 1-22.
[31] Ardanaz, M., S. Otálvaro-Ramírez and C. Scartascini (2023), “Does information about citizen participation initiatives increase political trust?”, World Development, Vol. Volume 162/106132, https://doi.org/10.1016/j.worlddev.2022.106132.
[83] Argyle, L. et al. (2023), Leveraging AI for democratic discourse: Chat interventions can improve online political conversations at scale, https://doi.org/10.1073/pnas.2311627120.
[148] Arunachalam, R. and S. Sarkar (2013), The New Eye of Government: Citizen Sentiment Analysis in Social Media.
[11] Asseh, B. et al. (2025), Et si l’IA était au service de la démocratie ? L’exemple du grand débat national, https://www.jean-jaures.org/publication/et-si-lia-etait-au-service-de-la-democratie-lexemple-du-grand-debat-national/.
[54] Belgian Presidency (2024), A Citizen’s View of Artificial Intelligence within the EU, https://demnet.hu/wp-content/uploads/2024/05/rapport-ia-en-web-v2.pdf.
[181] Benade, G., P. Gölz and A. Procaccia (2019), No Stratification Without Representation, https://doi.org/10.1145/3328526.3329578.
[12] Berditchevskaia, A. and P. Baeck (2020), The Future of Minds and Machines: How artificial intelligence can enhance collective intelligence.
[94] Berigüete, F. (2024), “Digital Revolution: Emerging Technologies for Enhancing Citizen Engagement in Urban and Environmental Management”, https://doi.org/10.3390/land13111921.
[137] Berryhill, J. et al. (2019), “Hello, World: Artificial intelligence and its use in the public sector”, OECD Working Papers on Public Governance, No. 36, OECD Publishing, Paris, https://doi.org/10.1787/726fd39d-en.
[173] Bertelsmann Stiftung (2025), Public AI Whitepaper, https://doi.org/10.11586/2025040.
[170] Bertelsmann Stiftung (2024), Forum Against Fakes: Citizens’ report on how to deal with disinformation, https://doi.org/10.11586/2024149.
[5] Bertelsmann Stiftung (2022), Under Construction: Citizen Participation in the European Union, https://www.bertelsmann-stiftung.de/de/publikationen/publikation/did/under-construction.
[165] Bertelsmann Stiftung (2018), Mitreden, Mitgestalten, Mitentscheiden: Five ideas for renewing democratic participation, https://www.bertelsmann-stiftung.de/de/publikationen/publikation/did/mitreden-mitgestalten-mitentscheiden-1.
[135] Bjarnason, R., D. Gambrell and J. Lanthier-Welch (2024), “Using Artificial Intelligence to Accelerate Collective Intelligence - Policy Synth and Smarter Crowdsourcing”, https://doi.org/10.48500/arXiv.2407.13960.
[52] Brasil Participativo (2025), Brasil Participativo, https://www.gov.br/pt-br/participacao-social.
[145] Bussola Tech and Luís Kimaid (2024), Integrating Artificial Intelligence to Legislative Services - Ulysses Suite in the Chamber of Deputies of Brazil, https://library.bussola-tech.co/p/ulysses-chamber-deputies-brazil.
[78] Byers, P., R. Zeller and B. Byers (2021), “Focus Groups Methods”, https://doi.org/10.4324/9781410604545-12.
[17] Campagnucci, F. et al. (2024), Artificial Intelligence for Participation.
[139] Cardoso, A. (2023), “Do we need a Chat-GPT-Gov? The importance of technology for effective access to public information.”, https://doi.org/10.2139/ssrn.4365773.
[9] Carson, L. and S. Elstub (2019), Carson, L., & Elstub, S. (2019). Comparing Participatory and Deliberative Democracy, https://www.newdemocracy.com.au/wp-content/uploads/2019/04/RD-Note-Comparing-Participatory-and-Deliberative-Democracy.pdf.
[48] CESE (2024), Demander Comprendre la Convention Citoyenne sur la Fin de Vie, https://panoramic.make.org/customer/cese/event/convention-citoyenne-sur-la-fin-de-vie?displayonboarding=false.
[13] Chemier, B. and J. Hotsko (2024), Can Artificial Intelligence Drive Citizen Engagement to Address Climate Change?, https://www.undp.org/acceleratorlabs/blog/visualizing-future-artificial-intelligence-climate-action.
[125] Chwalisz, C. and S. McKinney (2025), Five dimensions of scaling democratic deliberation: With and beyond AI, https://www.demnext.org/projects/five-dimensions-of-scaling-democratic-deliberation-with-and-beyond-ai.
[100] City of Helsinki (2025), Kallio summer streets, https://www.hel.fi/en/urban-environment-and-traffic/urban-planning-and-construction/plans-and-building-projects/kallio-summer-streets.
[51] Civic Engagement Commission (2025), Participate NYC, https://www.participate.nyc.gov/.
[157] Civic Tech Field Guide (2023), Prompts for Change: 110 ways civil society orgs can use AI, https://civictech.guide/prompts-for-change/.
[45] Combaz, A. et al. (2024), Applications of Artificial Intelligence Tools to.
[46] Consul Democracy (2025), New project: Open-source, LLM-powered Civic Assistant, https://consuldemocracy.org/2025/03/announcing-ai-assistant-project/.
[130] Consul Democracy Foundation (2025), New project: Open-source, LLM-powered Civic Assistant, https://consuldemocracy.org/2025/03/announcing-ai-assistant-project/.
[90] Co-operation Ireland (2025), Voices Matters, https://cooperationireland.org/projects/voice-matters/.
[57] Cortés-Cediel, M. (2023), “Trends and challenges of e-government chatbots: Advances in exploring open government data and citizen participation content.”, Government Information Quarterly, Vol. 40/4, p. 101877, https://doi.org/10.1016/j.giq.2023.101877.
[79] Cortini, M., T. Galanti and S. Fantinelli (2019), “Focus Group Discussion: how many Participants in a Group?”, https://doi.org/10.6092/ISSN.1825-8670/9603.
[177] Dane, G. et al. (2024), “valuating a new framework for the participatory co-design of healthy public spaces using immersive virtual reality”, Computers, Environment and Urban Systems Volume 114, https://doi.org/10.1016/j.compenvurbsys.2024.102194.
[67] Dasgupta, A. et al. (2012), Impact of spam exposure on user engagement, https://doi.org/10.5555/2362793.2362796.
[62] De Gregorio, G. (2020), “Democratising online content moderation: A constitutional framework”, Computer Law & Security Review, Vol. 36, https://doi.org/10.1016/j.clsr.2019.105374.
[59] Decide_Madrid (2021), Como utilizar Decide Madrid y su asistente virtual “Clara”, https://decide.madrid.es/system/documents/attachments/000/007/952/original/3779c5bafc46109e4b9f57fbeecf680b5224c302.pdf.
[91] Dembrane (2025), Derry/Londonderry made a blueprint for a safer, more inclusive city using Dembrane, https://www.dembrane.com/en-US/blog/londonderry-case-study.
[129] DemocracyNext (2024), Tech-Enhanced Deliberative Assemblies, https://www.demnext.org/projects/tech-enhanced-citizens-assemblies.
[60] Demokratie Today (2025), , https://demokratie.today/.
[63] Dias Oliva, T., D. Antonialli and A. Gomes (2021), “Fighting Hate Speech, Silencing Drag Queens? Artifcial Intelligence in Content Moderation and Risks to LGBTQ Voices Online”, Security & Culture, Vol. 25/2, pp. 700-732, https://doi.org/10.1007/s12119-020-09790-w.
[69] Dillard, K. (2013), “Envisioning the Role of Facilitation in Public Deliberation”, Journal of Applied Communication Research, Vol. 41/3, https://doi.org/10.1080/00909882.2013.826813.
[22] Dos Santos, L., E. Saltz and T. Acosta (2025), Using LLMs to support online communities, https://medium.com/jigsaw/using-llms-to-support-online-communities-03e2dcc7b451.
[18] Duberry, J. (2022), Artificial Intelligence and Democracy, https://library.oapen.org/bitstream/handle/20.500.12657/91687/210733.pdf?sequence=1&isAllowed=y.
[161] Dwyer, A. (2024), Enabling Secure Democratic Ecosystems through AI, https://static.ie.edu/CGC/AI4D%20Paper%204%20Enabling%20Secure%20Democratic%20Ecosystems.pdf.
[117] Eerola, E. et al. (2025), Resident panel and Polis experiments in the wellbeing services counties in 2023–2024, https://www.sitra.fi/julkaisut/asukaspaneeli-ja-polis-kokeilut-hyvinvointialueilla-2023-2024/.
[122] Ehsassi, M. (2024), Activated Citizenship: The Transformative Power of Citizens’ Assemblies, Routledge.
[49] European Commission (2021), Conference on the Future of Europe, https://wayback.archive-it.org/12090/20230418091815/https:/futureu.europa.eu/.
[92] European Committee of the Regions (2025), Becoming a citizen engagement expert - YEP Winter School 2025: Kick-off conference, https://cor.europa.eu/en/plenaries-events/becoming-citizen-engagement-expert-yep-winter-school-2025-kick-conference.
[143] European Parliament (2025), Ask the EP Archives, https://archidash.europarl.europa.eu/ep-archives-anonymous-dashboard.
[50] Fabbrini, F. (2021), “The Conference on the Future of Europe: Process and prospects”, Review of European Law in Context, https://doi.org/10.1111/eulj.12401.
[40] Fillet, S. (2024), Cambridge City Council analyzes input 50% faster with Go Vocal’s AI assistant, https://www.govocal.com/case-studies/cambridge-city-council-analyzes-input-50-faster-with-go-vocals-ai-assistant.
[88] Fishkin, J. et al. (2021), Final Report - LXS400 Chile Delibera, https://deliberation.stanford.edu/news/lxs-400-national-deliberative-poll-chile.
[101] Fitkov-Norris, E. and N. Kocheva (2025), Leveraging AI for strategic foresight: Unveiling future, Edward Elgar Publishing, https://doi.org/10.4337/9781035310586.00023.
[61] Flew, T., F. Martin and N. Suzor (2019), “Internet regulation as media policy: Rethinking the question of digital communication platform governance.”, Journal of Digital Media & Policy 10, pp. 33-50, https://doi.org/10.1386/jdmp.10.1.33_1.
[37] French Government (2019), Grand Débat, https://granddebat.fr/.
[73] Fung, A. (2005), “Deliberation before the revolution: Toward an ethics of deliberative democracy in an unjust world”, Political Theory, Vol. 33/3, pp. 397-419.
[14] García, D. and E. Al. (2023), Democracy Technologies in Europe: Online Participation, Deliberation and Voting, Innovation in Politics Institute, https://www.idea.int/publications/catalogue/democracy-technologies-europe.
[182] Gasiorowska, A. (2023), “Sortition and its Principles: Evaluation of the Selection Processes of Citizens’ Assemblies”, Journal of Deliberative Democracy, Vol. 19/1, https://doi.org/10.16997/jdd.1310.
[142] Gast Romps, N. (2023), Searching for Solutions: MITRE Tool Simplifies Freedom of Information Act Requests, https://www.mitre.org/news-insights/impact-story/mitre-tool-simplifies-freedom-information-act-requests.
[33] GCS (2024), GCS generative AI policy, https://gcs.civilservice.gov.uk/publications/gcs-generative-ai-policy/.
[64] Gillespie, T. (2018), Custodians of the Internet. Platforms, Content Moderation, and the Hidden Decisions That Shape Social Media, Yale University Press, https://yalebooks.yale.edu/book/9780300261431/custodians-of-the-internet/.
[58] Gobierno Abierto Bogotá (2025), , https://gobiernoabiertobogota.gov.co/transparencia.
[15] Goldberg, B. and A. al. (2024), “AI and the Future of Digital Public Squares”, https://arxiv.org/abs/2412.09988.
[41] Government of Finland (2025), National Dialogues, https://kansallisetdialogit.fi/en/.
[132] Government Technology Agency Singapore (2025), “What is an agent?”, Agentic AI primer, https://playbooks.aip.gov.sg/agentic-ai-primer/02_what_is_an_agent/.
[140] Gregory, M. et al. (2024), The findings of our first generative AI experiment: GOV.UK Chat, https://insidegovuk.blog.gov.uk/2024/01/18/the-findings-of-our-first-generative-ai-experiment-gov-uk-chat/.
[183] Griffin, J. et al. (2015), “Understanding Participant Representativeness in Deliberative Events: A Case Study Comparing Probability and Non-Probability Recruitment Strategies.”, Journal of Deliberative Democracy, Vol. 11/4, https://doi.org/10.16997/jdd.221.
[96] Guridi, J. et al. (2024), “Image Generative AI to Design Public Spaces: A Reflection of How AI Could Improve Co-Design of Public Parks”, Digital Government: Research and Practice, https://doi.org/10.1145/3656588.
[23] Hadfi, R. et al. (2021), “Argumentative Conversational Agents for Online Discussions”, Journal of Systems Science and Systems Engineering, https://doi.org/10.1007/s11518-021-5497-1.
[86] Hadfi, R. et al. (2021), “Argumentative Conversational Agents for Online Discussions”, Journal of Systems Science and Systems Engineering, https://doi.org/10.1007/S11518-021-5497-1.
[128] Healthy Democracy (2024), Deschutes Civic Assembly on Youth Homelessness, https://healthydemocracy.org/home/projects/deschutescivicassembly/.
[133] Iason, G. et al. (2025), We need a new ethics for a world of AI agents, https://www.nature.com/articles/d41586-025-02454-5.
[131] Ilves, L. et al. (2025), The Agentic State - How Agentic AI Will Revamp 10 Functional Layers of Public Administration.
[141] INPS (2025), INPS.
[87] Ito, T., R. Hadfi and S. Suzuki (2021), “An Agent that Facilitates Crowd Discussion”, https://doi.org/10.1007/s10726-021-09765-8.
[160] Juneja, P. (2024), Artificial Intelligence for Electoral Management, https://doi.org/10.31752/idea.2024.31.
[85] Kim, S. et al. (2021), Moderator Chatbot for Deliberative Discussion: Effects of Discussion Structure and Discussant, https://doi.org/10.1145/3449161.
[104] Kittredge, A. (2024), BetaNYC presents FloodGen, an advocacy tool that uses generative AI imagery, https://www.beta.nyc/2024/05/03/betanyc-presents-floodgen-a-advocacy-tool-that-uses-generative-ai-imagery/https://www.beta.nyc/2024/05/03/betanyc-presents-floodgen-a-advocacy-tool-that-uses-generative-ai-imagery/.
[147] Kumi, S. et al. (2024), “Uncovering Concerns of Citizens Through Machine Learning and Social Network Sentiment Analysis”, https://doi.org/10.1109/ACCESS.2024.3426329.
[81] Landemore, H. (2022), “Landemore, H. (2022). Can AI bring deliberative democracy to the masses?”, HAI Weekley Seminar, pp. 1-35.
[154] Linares, C. (2023), Periodistas de México y Brasil desarrollan herramienta de inteligencia artificial para detectar discurso de odio en línea contra comunicadores, https://latamjournalismreview.org/es/articles/periodistas-de-mexico-y-brasil-desarrollan-herramienta-de-inteligencia-artificial-para-detectar-discurso-de-odio-en-linea-contra-comunicadores/.
[3] Loisel, M. and N. Rio (2024), Pour en finir avec la démocratie participative, https://www.editionstextuel.com/livre/pour_en_finir_avec-_la_democratie_participative.
[32] Lovari, A. and F. De Rosa (2025), “Exploring the Challenges of Generative AI on Public Sector Communication in Europe”, Media and Communication 13.
[115] Lovio, I. (2025), Sitra to significantly increase funding for solutions that renew Finland, https://www.sitra.fi/en/news/sitra-to-significantly-increase-funding-for-solutions-that-renew-finland/.
[53] Lugano Living Lab (2019), , https://www.luganolivinglab.ch/ricerca/project-debater---speech-by-crowd.
[120] Makinson, L. et al. (2025), Consult Evaluation: Scottish Government’s Non-surgical cosmetic procedures consultation, https://ai.gov.uk/docs/Scot_Gov_NSCP_Evaluation_Report.pdf.
[44] MAPLE (2025), MAPLE - Massachussetts Platform for Legislative Engagement, https://www.mapletestimony.org/.
[156] Masood Alavi, D. (2022), “Using Artificial Intelligence for Peacebuilding”, Journal of Peacebuilding & Development, Vol. 17/2, https://doi.org/10.1177/15423166221102757.
[16] McKinney, S. (2024), “Integrating Artificial Intelligence Into Citizens’”, Journal of Deliberative Democracy, Vol. 20/1, pp. 1-12, https://doi.org/10.16997/jdd.1556.
[70] Moore, A. (2012), “Following from the front: theorizing deliberative facilitation”, Critical Policy Studies, Vol. 6, https://doi.org/10.1080/19460171.2012.689735.
[25] Morozov, E. (2013), To save everything, click here: The folly of technological solutionism, PublicAffairs.
[153] Nesta (2021), Collective crisis intelligence for frontline humanitarian response, https://media.nesta.org.uk/documents/Nesta_Collective_crisis_intelligence_for_frontline_humanitarian_response_Final.pdf.
[123] Niemeyer, S. (2014), Scaling up deliberation to mass publics: Harnessing mini-publics in a deliberative system.
[107] OECD (2025), An immersive technologies policy primer, https://doi.org/10.1787/cf39863d-en.
[4] OECD (2025), Exploring New Frontiers in Citizen Participation in the Policy Cycle, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/77f5098c-en.
[146] OECD (2025), Getting the Public on Side: How to Make Reforms Acceptable by Design, OECD Publishing, https://doi.org/10.1787/262255fd-en.
[28] OECD (2025), Governing with Artificial Intelligence: The state of play and way forward in core government functions, OECD Publishing.
[2] OECD (2025), “Tackling civic participation challenges with emerging technologies: Beyond the hype”, OECD Public Governance Policy Papers, No. 72, OECD Publishing, Paris, https://doi.org/10.1787/ec2ca9a2-en.
[47] OECD (2025), “Tackling civic participation challenges with emerging technologies: Beyond the hype”, OECD Public Governance Policy Papers, No. 72, OECD Publishing, Paris, https://doi.org/10.1787/ec2ca9a2-en.
[162] OECD (2024), “Exploring New Frontiers of Citizen Participation in the Policy Cycle”, https://www.oecd.org/content/dam/oecd/en/about/programmes/reinforcing-democracy-initiative/Exploring-New-Frontiers-of-Citizen-Participation-Discussion-Paper.pdf.
[112] OECD (2024), Facts not Fakes: Tackling Disinformation, Strengthening Information Integrity, OECD Publishing, Paris, https://doi.org/10.1787/d909ff7a-en.
[171] OECD (2024), “Framework for Anticipatory Governance of Emerging Technologies”, OECD Science, Technology and Industry Policy Papers, No. 165, OECD Publishing, Paris, https://doi.org/10.1787/0248ead5-en.
[151] OECD (2024), Global Trends in Government Innovation 2024: Fostering Human-Centred Public Services, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/c1bc19c3-en.
[175] OECD (2024), “Governing with Artificial Intelligence: Are governments ready?”, OECD Artificial Intelligence Papers, No. 20, OECD Publishing, Paris, https://doi.org/10.1787/26324bc2-en.
[178] OECD (2024), OECD Artificial Intelligence Review of Germany, https://doi.org/10.1787/609808d6-en.
[7] OECD (2024), OECD Survey on Drivers of Trust in Public Institutions – 2024 Results: Building Trust in a Complex Policy Environment, OECD Publishing, Paris, https://doi.org/10.1787/9a20554b-en.
[152] OECD (2024), Practical Guide for Policymakers on Protecting and Promoting Civic Space, OECD Publishing, https://doi.org/10.1787/6c908b48-en.
[1] OECD (2024), The OECD Reinforcing Democracy Initiative: Monitoring Report – Assessing Progress and Charting the Way Forward, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/9543bcfb-en.
[93] OECD (2023), Global Trends in Government Innovation 2023, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/0655b570-en.
[167] OECD (2023), OECD Database of Representative Deliberative Processes, https://airtable.com/appP4czQlAU1My2M3/shrX048tmQLl8yzdc/tblrttW98WGpdnX3Y/viwX5ZutDDGdDMEep.
[6] OECD (2023), Open Government for Stronger Democracies: A Global Assessment, OECD Publishing, https://doi.org/10.1787/5478db5b-en.
[164] OECD (2022), Building Trust to Reinforce Democracy: Main Findings from the 2021 OECD Survey on Drivers of Trust in Public Institutions, Building Trust in Public Institutions, OECD Publishing, Paris, https://doi.org/10.1787/b407f99c-en.
[29] OECD (2022), “OECD Framework for the Classification of AI systems”, OECD Digital Economy Papers, No. 323, OECD Publishing, Paris, https://doi.org/10.1787/cb6d9eca-en.
[8] OECD (2022), OECD Guidelines for Citizen Participation Processes, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/f765caf6-en.
[27] OECD (2022), The Protection and Promotion of Civic Space: Strengthening Alignment with International Standards and Guidance, OECD Publishing, Paris, https://doi.org/10.1787/d234e975-en.
[30] OECD (2021), OECD Report on Public Communication: The Global Context and the Way Forward, OECD Publishing, Paris, https://doi.org/10.1787/22f8031c-en.
[163] OECD (2020), “Innovative Citizen Participation and New Democratic Institutions: Catching the Deliberative Wave”, OECD Publishing, https://www.oecd.org/en/publications/innovative-citizen-participation-and-new-democratic-institutions_339306da-en.html.
[71] OECD (2020), Innovative Citizen Participation and New Democratic Institutions: Catching the Deliberative Wave, OECD Publishing, https://doi.org/10.1787/339306da-en.
[176] OECD (2019), Open Government Data, https://www.oecd.org/gov/digital-government/open-government-data.htm#:~:text=What%20is%20Open%20Government%20Data,quantities%20of%20data%20and%20information.
[136] OECD (2017), Recommendation of the Council on Open Government.
[166] OECD (2001), Citizens as Partners: Information, Consultation and Public Participation in Policy-Making, OECD Publishing, Paris, https://doi.org/10.1787/9789264195561-en.
[169] OECD (forthcoming), Civic Tech for Democracy.
[174] OECD/CAF (2022), The Strategic and Responsible Use of Artificial Intelligence in the Public, OECD Publishing, https://doi.org/10.1787/1f334543-en.
[179] OECD/UNESCO (2024), G7 Toolkit for Artificial Intelligence in the Public Sector, https://doi.org/10.1787/421c1244-en.
[68] Office of the President of the Slovak Republic (2023), TrollWallAI - Testimonials, https://www.trollwall.ai/case-studies.
[42] Open Government Partnership (2023), Strengthening National Dialogues (FI0036), https://www.opengovpartnership.org/members/finland/commitments/fi0035/.
[144] OpenCouncil (2025), OpenCouncil, https://opencouncil.gr/.
[95] Otherngrafen, F., L. Sievers and E. Reinecke (2024), “From Vision to Reality: The Use of Artificial Intelligence in Different Urban Planning Phases”, https://doi.org/10.17645/up.8576.
[111] Ovadya, A. (2023), “Bridging Systems: Open Problems for Countering Destructive Divisiveness across Ranking, Recommenders, and Governance”, Arxiv, https://arxiv.org/abs/2301.09976.
[109] Ovadya, A. (2023), “Generative CI’through Collective Response Systems”, https://doi.org/10.48550/arXiv.2302.00672.
[114] Paice, A. and M. Rausch (2022), Creating an online conversation between a nation and a mini-public: a case study on Polis & the Austrian Citizens’ Climate Council, https://uploads-ssl.webflow.com/65b77644e6021e9021de8916/65ddca5dfe2b472c6e1256f7_Case_Study_KLIMARAT.pdf.
[168] Participedia (2025), Participedia, https://participedia.net/.
[113] People Powered (2025), 2025 Rating for Digital Participation Tools, https://www.peoplepowered.org/platform-ratings.
[127] People Powered (2025), Guide to Digital Participation Platform, https://drive.google.com/file/d/1J0BlXA5d6Fcgbk7MIZTqaq36eVgl9jaP/view?pli=1.
[159] People’s Law School (2025), Beagle+, https://www.peopleslawschool.ca/about/beagle/.
[43] Perez, O. (2008), “Complexity, information overload, and online deliberation”, ISJLP 5.
[102] Polchar, J. and C. Karle (2023), Instant Scenarios: How to fast-forward your foresight and skip straight to strategy, https://oecd-opsi.org/blog/instant-scenarios/.
[97] Preussner, A. et al. (2025), UrbAI: Exploring the Possibilities of Generative AI Image Processing to Promote Citizen Participation, https://doi.org/10.1145/3706598.3713803.
[138] Proscovia, S. (2025), “Democratizing Open Data Through AI: The Case of Sweden”, Revue électronique suisse de science de l’information, https://doi.org/10.55790/journals/ressi.2025.e1823.
[105] Quan, S. (2022), “Urban-GAN: An artificial intelligence-aided computation system for plural urban design”, Environment and Planning B: Urban Analytics and City Science, Vol. 49/2, https://doi.org/10.1177/23998083221100550.
[65] Roberts, S. (2017), Content Moderation, Springer, https://doi.org/10.1007/978-3-319-32001-4_44-1.
[19] Romberg, J. and T. Escher (2024), “Making Sense of Citizens’ Input through Artificial Intelligence: A Review of Methods for Computational Text Analysis to Support the Evaluation of Contributions in Public Participatio”, Digital Government: Research and Practice, pp. 1-30, https://doi.org/10.1145/3603254.
[75] Romberts, J. and O. Escobar (2015), Involving communities in deliberation: A study of three citizens’ juries on onshore wind farms in Scotland..
[149] Rosenthal, A. (2019), When old technology meets new: how UN Global Pulse is using radio and AI to leave no voice behind, https://unfoundation.org/blog/post/when-old-technology-meets-new-how-un-global-pulse-is-using-radio-and-ai-to-leave-no-voice-behind/.
[121] Sarabi, R. (2025), How AI Can Unlock Public Wisdom and Revitalize Democratic Governance, https://carnegieendowment.org/posts/2025/07/how-ai-can-unlock-public-wisdom-and-revitalize-democratic-governance?lang=en.
[124] ScaleDem (2025), Theory of Scaling, https://scaledem.eu/insights/theory-of-scaling/.
[38] Schneider, B. and N. Sanders (2023), AI could shore up democracy - here’s one way, https://theconversation.com/ai-could-shore-up-democracy-heres-one-way-207278 (accessed on 23 May 2024).
[26] Selbst, A. et al. (2018), Fairness and Abstraction in Sociotechnical Systems, https://ssrn.com/abstract=3265913.
[89] Senato de la República de Chile (2021), Lanzan “Lxs 400¿Chile Delibera”: chilenos pensando políticas públicas, https://www.senado.cl/comunicaciones/noticias/lanzan-lxs-400chile-delibera-chilenos-pensando-politicas-publicas.
[116] Sitra (2023), FAQ - What is the ‘What do you think, Finland’ experiment all about?, https://www.sitra.fi/en/articles/faq-what-is-the-what-do-you-think-finland-experiment-all-about/.
[24] Small, C. et al. (2023), Opportunities and Risks of LLMs for Scalable Deliberation with Polis, https://doi.org/10.48550/arxiv.2306.11932.
[74] Smith, G. (2009), Democratic innovations: designing institutions for citizen participation, Cambridge University Press.
[20] Starke, C. and M. Lünich (2020), “Artificial intelligence for political decision-making in the European Union: Effects on citizens’ perceptions of input, throughput, and output legitimacy”, Data&Policy, https://doi.org/10.1017/dap.2020.19.
[76] Stromer-Galley, J. (2007), “Measuring Deliberation’s Content: A Coding Scheme”, Journal of Public Deliberation, Vol. 3/1, https://doi.org/10.16997/jdd.50.
[158] Talk to the City (2023), Heal Michigan, https://talktothecity.org/report/heal-michigan.
[77] Tang, K. and A. Davis (1995), “Critical factors in the determination of focus group size”, https://doi.org/10.1093/FAMPRA/12.4.474.
[82] Tessler, M. and A. al. (2024), “AI can help humans find common ground in democratic deliberation”, Science, Vol. 386, https://doi.org/10.1126/science.adq2852.
[180] Ubaldi, B. (2019), State of the art in the use of emerging technologies in the public sector, OECD Publishing, https://doi.org/10.1787/932780bc-en.
[99] UrbanistAI (2023), Co-Designing the Helsinki Summer Streets, https://site.urbanistai.com/helsinki-summer-streets-1.
[39] Valle-Cruz, D. et al. (2019), A Review of Artificial Intelligence in Government and its Potential from a Public Policy Perspective, Association for Computing Machinery, https://doi.org/10.1145/3325112.3325242.
[56] Van Noordt, C. and G. and Misuraca (2019), “New Wine in Old Bottles: Chatbots in Government - Exploring the transformative impact of Chatbots in public service delivery”, LECTURE NOTES IN COMPUTER SCIENCE, Vol. Electronic Participation - International Conference , 02-05 September 2019, San Benedetto del Tronto, Italy, pp. 49-59, https://doi.org/10.1007/978-3-030-27397-2_5.
[110] Verhulst, S. (2018), “Where and when AI and CI meet: exploring the intersection of artificial and collective intelligence towards the goal of innovating how we govern”, AI & Society, Vol. 33.
[172] Vocdoni (2022), First Public Institution in Spain Holds Vocdoni Referendum | Bellpuig Council, https://blog.vocdoni.io/referendum-bellpuig.
[150] Wade, A., K. Chahal and M. Melton (2025), Introducing Assist – the dynamic AI tool rapidly transforming government communications, https://gcs.civilservice.gov.uk/blog/introducing-assist-the-dynamic-ai-tool-rapidly-transforming-government-communications/.
[35] Weeks, B., A. Halversen and G. Neubaum (2024), “Too scared to share? Fear of social sanctions for political expression on social media”, Journal of Computer-Mediated Communication, Vol. 29/1.
[108] Weyl, E., A. Tang and C. Plurality (2023), Augmented Deliberation, https://www.plurality.net/v/chapters/5-4/eng/?mode=dark#fn21.
[103] Weyl, G., A. Tang and P. Community (2024), Immersive Shared Reality, https://www.plurality.net/v/chapters/5-2/eng/?mode=dark.
[80] White, K., N. Hunter and K. Greaves (2022), Facilitating Deliberation - A practical guide, MosaicLab.
[72] White, K., N. Hunter and K. Greaves (2022), Facilitating Deliberation: a Practical Guide.
[36] Wihbey, J. (2024), “AI and Epistemic Risk for Democracy: A Coming Crisis of Public Knowledge?”, https://doi.org/10.2139/ssrn.4805026.
[155] Wildi, D., M. Braizat and M. Evrard (2024), Leveraging Artificial Intelligence for the Pursuit of Human Rights, https://www.graduateinstitute.ch/sites/internet/files/2024-09/ARP50_Final-Report_Edit-27.07.2024---Daniela-Wildi.pdf.
[98] Williams, S. et al. (2024), “People-Powered Gen AI: Collaborating with Generative AI for Civic Engagement”, An MIT Exploration of Generative AI, https://doi.org/10.21428/e4baedd9.f78710e6.
[84] Wyss, D. and S. Beste (2017), “Artificial facilitation: Promoting collective reasoning within asynchronous discussions”, Journal of Information Technology & Politics, https://doi.org/10.1080/19331681.2017.1338175.
[106] Xu, H. et al. (2024), “Leveraging generative AI for urban digital twins: a scoping review on the autonomous generation of urban data, scenarios, designs, and 3D city models for smart city advancement”, https://doi.org/10.1007/s44212-024-00060-w.
[66] Yang, R. et al. (2021), Scalable Detection of Promotional Website Defacements in Black Hat SEO Campaigns, https://www.usenix.org/conference/usenixsecurity21/presentation/yang-ronghai.
[21] Zhang, A. et al. (2023), Deliberating with AI: Improving Decision-Making for the Future through Participatory AI Design and Stakeholder Deliberation, https://doi.org/10.1145/3579601.
[118] Zisengwe, M. and M. Stempeck (2025), What you should know about digital participation & AI, https://www.peoplepowered.org/news-content/what-governments-should-know-about-digital-participation-amp-ai.
← 1. The OECD understands representative deliberative processes as citizen participation processes involving a limited number of participants selected through stratified sortition (“civic lottery”) to represent the broader society. Participants are given time and means to learn about the policy issue at stake, hear from different perspective, form their own opinions and formulate recommendations based on informed deliberation. The extent to which stratified sortition succeeds in representing society is an open debate in the deliberation literature (Benade, Gölz and Procaccia, 2019[181]; Gasiorowska, 2023[182]; Griffin et al., 2015[183]).