Architectures of Digital Agency: A Framework for Enhancing Democratic Discourse and Collective Intelligence

The evolution of democratic governance in the twenty-first century is increasingly defined by the intersection of institutional design and computational innovation. As traditional representative systems face crises of legitimacy and efficiency, a diverse ecosystem of civic technologies has emerged to bridge the widening gap between citizens and decision-making power. These technologies are not merely functional tools; they represent a fundamental shift in digital civic infrastructure, designed to catalyze individual and collective agency through structured participation, deliberation, and knowledge synthesis [cite: 1, 2, 3]. This transition is underpinned by a move from “passive consumption” of political information to “active co-creation” within the public sphere, facilitated by modular architectures that allow for the scaling of democratic innovations [cite: 1, 4, 5, 6]. The rapid proliferation of artificial intelligence has further complicated this landscape, introducing agentic frameworks that assume roles as moderators, facilitators, and teachers, thereby reshaping the conditions under which human agency is enacted [cite: 7, 8, 9].

Taxonomy and Classification of the Digital Democratic Ecosystem

To navigate the complex landscape of digital democracy, it is essential to establish a rigorous taxonomy based on intent, target groups, and the specific democratic functions these tools are designed to support. The intent of technological interventions ranges from fostering informal conversation to achieving high-stakes policy consensus and conducting systemic research [cite: 9, 10, 11]. These interventions target a multi-layered group of stakeholders, including individual citizens, professional facilitators, moderators, and government administrators, each of whom requires distinct affordances to participate effectively [cite: 12, 13, 14].

Strategic Dimensions of Intent and Function

The primary classification dimensions for enhancing democratic discourse involve a matrix of intent and function. Intent refers to the desired outcome of the engagement, while the democratic function identifies the stage of the governance process where the tool is applied.

Intent Category	Functional Alignment	Primary Mechanism	Desired Outcome
Conversation	Inform, Listen	Threaded dialogue, instant messaging	Relationship building, information exchange
Consensus	Deliberate, Decide	Opinion mapping, clustering algorithms	Collective agreement, policy trade-offs
Participation	Decide, Govern	Modular workflows, participatory budgeting	Direct civic action, resource allocation
Knowledge	Inform, Learn	Open data portals, legislative databases	Civic literacy, evidence-based discourse
Fact-checking	Moderation, Learn	Crowdsourced verification, AI note writing	Information integrity, bias reduction
Research	Deliberate, Learn	AI synthesis, qualitative analysis	Pattern recognition, insight generation

The functional spectrum begins with “Informing” and “Listening,” which establish the knowledge base necessary for participation. This establishes a foundation of civic literacy and accessibility, allowing for more advanced functions such as “Deliberating” and “Deciding” [cite: 1, 4]. Deliberation involves the structured evaluated of arguments to achieve mutual understanding, while the decision phase translates these insights into implementable policy [cite: 15, 16]. Finally, the “Governing” and “Learning” functions ensure accountability through traceability and iterative feedback loops that allow both citizens and institutions to evolve their practices [cite: 4, 5, 17, 18].

Target Group Dynamics and Influence Factors

The effectiveness of these tools is significantly impacted by the specific needs of different target groups. For citizens, influence factors such as cognitive load, inclusion, and safety are paramount [cite: 9, 19]. Digital engagement is not merely about providing a platform; it is about ensuring that the platform is safe to inhabit and intelligible to use for vulnerable or historically marginalized groups [cite: 19]. For facilitators and community managers, the focus shifts toward tools that reduce administrative overhead and automate the synthesis of large-scale input, allowing them to prioritize the relational dimensions of community leadership [cite: 14, 20, 21].

Target Group	Primary Influence Factor	Key Requirement	Strategic Focus
Citizens	Cognitive Load, Inclusion	Accessibility, simplification, safety	Removing participation barriers
Facilitators	Civility, Reflection	Moderation automation, synthesis tools	Enhancing deliberative quality
Moderators	Transparency, Quality	Fact-checking, toxicity sensors	Protecting information integrity
Researchers	Reflection, Consensus	Advanced analytics, data synthesis	Discovering collective patterns
Governments	Accountability, Action	Traceability, implementation tracking	Legitimacy and responsiveness
Managers	Participation, Empathy	Activity scores, workflow automation	Sustaining long-term engagement

Modular Architectures for Participatory Democracy

The architectural design of civic platforms fundamentally shapes the nature of democratic interaction. Two dominant models have emerged: modular, general-purpose systems for broad participation and specialized platforms for consensus-building and opinion mapping.

Decidim and Consul: The Technopolitical Paradigm

Decidim and Consul are the leading open-source infrastructures for participatory democracy, used by over 500 institutions worldwide [cite: 3, 22]. These platforms are characterized by their “technopolitical” philosophy, which posits that the design choices and political implications of software are inextricably linked [cite: 3, 5]. Decidim, born out of the 15M movement in Barcelona, is built as a modular framework that allows organizations to construct custom democratic spaces using a combination of “participatory spaces” (e.g., assemblies, processes) and “participatory components” (e.g., proposals, voting, accountability) [cite: 5, 22].

This modularity allows Decidim to operationalize key deliberative principles, including inclusion, reason-giving, transparency, and traceability [cite: 17]. Unlike unstructured social media environments, Decidim structures interaction across identifiable stages—from proposal submission and public discussion to amendment, prioritization, and institutional response [cite: 17, 23]. This sequential organization transforms participation from a continuous stream of reactive comments into a collective process oriented towards tangible outcomes [cite: 17].

Consul Democracy, originally developed in Madrid, focuses on a more streamlined set of participatory functions tailored for municipal authorities [cite: 22]. It provides a mature system for citizen proposals with digital signature collection, participatory budgeting, and local referenda [cite: 24]. Consul’s institutional role system is particularly well-adapted to local administration, ensuring that participants are verified and that the process adheres to strict compliance and accessibility standards [cite: 22].

Comparisons of Participatory Infrastructures

The choice between these platforms often rests on the specific democratic model an organization wishes to promote.

Architectural Dimension	Decidim	Consul Democracy
System Philosophy	Flexible framework for any democratic space	Structured tool for municipal authorities
Core Modular Units	Spaces (Processes, Assemblies, Conferences)	Functional Areas (Proposals, Budgets, Voting)
Technical Maturity	High modularity, complex recombinability	Mature institutional roles and local governance
Governance Model	Association-led (community and public funds)	Historically city-council led (Madrid)
Primary Goal	Creating diverse, custom democratic designs	Efficient implementation of 4 specific models
Transparency	High traceability via accountability component	Transparent signature and voting trails

Decidim’s “accountability” component is a unique feature that allows citizens to monitor whether and how a project is executed, establishing a direct link between public input and institutional action [cite: 5]. This traceability strengthens the legitimacy of the process, as citizens can evaluate responsiveness and understand constraints through reasoned justification rather than suspicion [cite: 17].

Advanced Deliberation and Consensus Engineering

While platforms like Decidim focus on participation and budgeting, other tools are specifically engineered for deep deliberation and the discovery of collective consensus in large, diverse groups.

Pol.is and the vTaiwan Model

Pol.is represents a significant innovation in opinion mapping and large-scale deliberation [cite: 24, 25]. Rather than allowing traditional threaded replies, which often degrade into mutual trolling, Pol.is restricts participants to “Agree,” “Disagree,” or “Pass” on discrete statements [cite: 26, 27]. This design nudges participants toward greater understanding of alternative viewpoints by strategically presenting statements from different opinion camps [cite: 28].

The underlying algorithm uses unsupervised machine learning to find the latent structure of the conversation, visualizing it as an “opinion landscape” where participants are clustered based on their shared perspectives [cite: 24, 29]. This enables the identification of “rough consensus”—statements that find agreement across different clusters—which often represent the most viable policy trade-offs [cite: 28, 30].

The vTaiwan process integrated Pol.is into a four-stage deliberative workflow:

1. Brainstorm Stage: Organizers and participants define the issue and write initial problem statements.
2. Preference Expression Stage: The public votes on these statements via Pol.is, creating an opinion landscape.
3. Deliberation Stage: Small-group, face-to-face (or hybrid) consultations are held, grounded in the data from the opinion mapping round.
4. Implementation Stage: Consensus items are ratified into new regulations or legislation [cite: 27, 30, 31].

This systematic approach has led to decisive government action in 80% of its cases, resolving disputes over complex issues like UberX regulation, the FinTech sandbox, and online alcohol sales [cite: 26, 27, 30].

AI-Enabled Deliberation and the Habermas Machine

The integration of Large Language Models (LLMs) has introduced “AI-mediated deliberation,” which aims to scale the quality of small-group discussion to thousands or millions of participants [cite: 16, 32, 33]. The “Habermas Machine” (HM), developed by researchers at DeepMind, is a flagship example of this trend [cite: 33]. The HM functions by synthesizing individual viewpoints into “consensus statements” intended for approval by all parties [cite: 33]. In empirical tests, the HM produced high-quality statements expressing common ground that garnered higher levels of endorsement than human-mediated statements, doing so in seconds rather than minutes [cite: 33].

These AI-facilitated platforms, such as deliberation.io, combine modular tools including Socratic dialogue, real-time preference visualization, and policy synthesis [cite: 16, 34]. By reducing participation costs while optimizing for inclusivity and consensus, these tools offer a scalable solution to the “representativeness deficit” that has historically limited deliberative mini-publics to small, hand-selected samples [cite: 16, 32].

AI as a Strategic Democratic Partner

In the contemporary civic tech landscape, artificial intelligence assumes several distinct roles, each contributing to the enhancement of individual and collective agency.

The AI Facilitator and Moderator

AI facilitators like those used in Listen Labs or Remesh allow for “qualitative research at scale” [cite: 10, 35]. These tools can talk to hundreds or thousands of participants simultaneously in real-time, using AI to probe for deeper meaning, analyze micro-expressions, and surface minority opinion clusters that traditional quantitative research might miss [cite: 10, 35, 36]. For community managers, this reduces the “backlog” of research and allows them to identify pattern-driven insights from community conversations without manual coding [cite: 14, 35].

AI moderators assume the role of filtering toxicity and enforcing civility [cite: 19, 20]. Tools like Normsy intervene in high-visibility toxic conversations on social media by helping users craft prosocial, pro-democratic responses, thereby expanding civic engagement into the broader digital environment [cite: 37]. This role is critical for protecting the “epistemic truth” and relationships in the face of algorithmic personalization [cite: 7].

The AI Community Coach and Teacher

The “Public Engagement Coach,” developed by The GovLab, represents the role of “Community Coach” [cite: 21]. This multi-agent tool guides public professionals through the planning of real engagement projects, asking targeted questions about goals, participants, and outreach strategies [cite: 18, 21]. By drawing from a curated library of engagement frameworks and case studies, the coach ensures that practitioners move from theoretical knowledge to practical, action-oriented project plans [cite: 21].

As a “Teacher,” AI-enabled platforms offer students and citizens a social-risk-free environment to practice democratic discourse [cite: 38, 39]. AI agents can be scripted to adopt ideological stances opposite to the user, ensuring that participants encounter diverse perspectives even in homogeneous cohorts [cite: 38]. This repetitive practice builds the “civic muscle” and confidence required for real-life political deliberation [cite: 39, 40].

AI Role	Democratic Function	Influence Factor	Key Implementations
Moderator	Listen, Govern	Civility, Toxicity reduction	Normsy, ITHACA Toxicity Sensors
Facilitator	Deliberate, Decide	Consensus, Inclusion	deliberation.io, Habermas Machine
Research Assistant	Learn, Inform	Information Quality, Quality	Listen Labs, Kimi Deep Research
Fact-checker	Inform, Moderation	Transparency, Accuracy	X AI Note Writer, Cofacts Chatbot
Community Coach	Govern, Decide	Reflection, Participation	GovLab AI Coach, Reboot Democracy
Teacher	Learn, Deliberate	Empathy, Polarization reduction	BYU Skill-training Agents
Deliberation Partner	Deliberate, Learn	Reflection, Intellectual humility	Stanford Online Deliberation, Minds

Information Integrity and Crowdsourced Fact-Checking

A fundamental threat to collective agency is the manipulation of the information environment [cite: 41, 42]. Civic tech has responded with a range of fact-checking and information quality tools that distribute the responsibility for verification across broad networks of contributors.

Community-Based Verification Systems

Platforms like X’s “Community Notes” (CN), formerly Birdwatch, rely on the “wisdom of crowds” to flag misleading content [cite: 11, 43]. CN requires contributors with different rating histories to agree that a note is helpful before it becomes visible, aiming for neutral, consensus-based judgments [cite: 11]. This design is being replicated across the industry, with Meta implementing similar voluntary user-driven moderation systems on Facebook, Instagram, and Threads [cite: 43].

In Taiwan, “Cofacts” operates as an open-source, citizen-driven platform that integrates a chatbot with LINE, the region’s most popular messaging app [cite: 44]. This allows over 300,000 users to verify suspicious messages through a network of 2,000 volunteering editors who have debunked over 87,000 articles [cite: 44]. This crowdsourced data is then made open to researchers and journalists to identify patterns of misinformation [cite: 44].

The Integration of AI in Information Verification

AI is increasingly used to complement human fact-checkers through tasks such as claim detection and “check-worthiness” estimation [cite: 11, 45]. The “AI Note Writer” on X uses LLMs to help draft notes based on information retrieval, while users continue to rate their helpfulness [cite: 11]. However, “thinking outside the bunk” suggests that technical solutions like debunking should be paired with underdeveloped approaches that focus on the social conditions of trust and institutional legitimacy [cite: 46].

Verification Tool	Mechanism	Focus	Target Group
Community Notes	Consensus-based crowdsourcing	Misinformation on X/Social Media	Citizens, Moderators
Cofacts	Chatbot + Editorial gatherings	Disinformation on Messaging apps	Citizens, Researchers
Meedan Check	Collaborative verification tiplines	Elections, Public health, Protests	Journalists, Civil Society
ClaimBuster	NLP and Supervised learning	Live fact-checking of public claims	Journalists, Researchers
Botometer	ML classification (Bot vs. Human)	Detecting automated manipulation	Researchers, Citizens

Enhancing Individual Agency and Reducing Cognitive Load

The core of the “agency problem” in digital democracy is the potential for technology to automate away human judgment [cite: 7, 47, 48]. Resilient civic infrastructure must therefore prioritize the preservation of human agency and the reduction of cognitive load for participants.

Digital Agency and Cognitive Scaffolding

Research identifies two forms of “proxy agency” with GenAI: “performance proxy agency,” where the tool does the task, and “learning proxy agency,” where the tool acts as a scaffold to expand the user’s capabilities [cite: 49, 50]. In democratic contexts, AI tools for simplification and translation are essential for inclusive participation among vulnerable groups [cite: 9, 19]. For instance, participants in vTaiwan found that AI helped them articulate opinions on social issues by “smoothing oral expression” and “reorganizing fragmented thoughts” [cite: 28].

However, there is a “backlash” against AI perceived as extractive or disempowering [cite: 40]. Effective design must include “meaningful frictions”—intentional elements that slow down interaction to encourage reflection rather than the “technological shortcuts” that hollow out democratic capacity [cite: 40, 51].

Polarization Reduction and Empathy Training

One of the most promising applications of AI in democracy is the active reduction of affective polarization. Experimental evidence shows that human-AI conversations can promote intellectual humility and a greater willingness to listen to political opponents [cite: 52, 53, 54, 55].

Johannes Walter’s 2025 research found that AI depolarization chatbots successfully persuaded participants to adopt more moderate positions, reducing overall ideological polarization by 20 percentage points [cite: 53, 56]. This effect was driven by learning and trust, as participants felt their individual concerns were better addressed by the AI than by traditional channels of persuasion [cite: 53, 56].

Influence Factor	AI Mechanism	Impact on Individual/Collective
Polarization Reduction	Counterarguing + Receptiveness	Increases intellectual humility and moderation
Cognitive Load	Simplification and Summarization	Enhances inclusion for vulnerable groups
Empathy	Serious games, avatars, persona simulation	Fosters social perspective-taking
Civility	Toxicity sensors and prosocial nudging	Reduces social risk and self-censorship
Transparency	Traceability and audit trails	Builds institutional trust and accountability
Participation	Workflow automation and invitations	Sustains habit-building and habitat safety

Technical Integration and Civic Data Synthesis

The “knowledge” function of democracy requires connecting disparate data sources to reveal relationships and patterns in the policy-making process [cite: 57].

Open Data and the Model Context Protocol (MCP)

“Civic AI Tools” connects AI assistants directly to government open data portals using the Model Context Protocol (MCP) [cite: 58, 59]. This framework allows users to explore complex datasets—such as housing violations or population demographics—using plain English [cite: 59]. By integrating official statistics directly into AI workflows, the system lowers the barriers to evidence-based advocacy for journalists, students, and civic technologists [cite: 58, 59].

Other platforms, such as “Digital Democracy” by CalMatters, combine transcripts of every word uttered in public hearings with bill information, financial contributions, and district data [cite: 57]. This database allows users to track the policy-making process and identify aberrations, such as a company representative’s testimony being linked directly to their campaign donations to the presiding legislators [cite: 57].

AI Research Assistants for Public Discourse

Community managers and researchers utilize a specialized ecosystem of AI research tools to synthesize public input and analyze discourse.

Research Tool	Primary Strength	Use Case
Kimi Deep Research	Comprehensive multi-source synthesis	Analyzing renewable energy policy trends
Claude Deep Research	Long-form writing, document analysis	Academic policy analysis, thesis writing
Gemini Deep Research	Multi-modal (text, image, video)	Visualizing neighborhood development
Perplexity	Real-time web search with citations	Fast literature discovery and validation
Elicit	Literature review automation	Systematic reviews for researchers
ScholarAI	Academic source support	Finding evidence for scholarly claims
MAXQDA / NVivo	Deep thematic coding and visual mapping	Qualitative discourse analysis

These tools enable the transition from “data collection” to “insight generation.” While basic survey tools produce dashboards, collective intelligence platforms must turn thousands of responses into decisions by identifying ranked themes and confidence scores [cite: 10].

Global Scenarios and the Digital Divide

The future of digital democracy is characterized by four potential “realist science fiction scenarios” that diverge along two axes: societal organization (authoritarianism vs. democracy) and mode of governance (representation vs. participation) [cite: 60].

1. Representative Digital Democracy: Traditional systems upgraded with digital voting and transparency portals [cite: 60, 61].
2. Participatory Digital Democracy: Radical models like Decidim or vTaiwan where citizens co-create policy [cite: 6, 60].
3. Representative Digital Authoritarianism: Use of surveillance and micro-targeting to maintain elite control [cite: 60, 62].
4. Participatory Digital Authoritarianism: Engagement used as a tool for “manufacturing consent” or identifying dissidents [cite: 60].

The UN E-Government Survey 2024 highlights that while global averages for digital government development (EGDI) have significantly improved, a persistent “digital divide” remains [cite: 63, 64]. The proportion of the population lagging in digital government development decreased from 45% to 22.4% between 2022 and 2024, yet Africa and Oceania continue to face challenges in infrastructure and capacity [cite: 63, 64].

Practical Frameworks for Practitioners and Facilitators

To move from “learning to doing,” practitioners require frameworks that bridge the “knowing-doing gap” in public engagement [cite: 18, 21].

The Spectrum of Community Engagement to Ownership

The “Spectrum of Community Engagement to Community Ownership” provides a relationship-mapping tool for assessing the depth of participation [cite: 65]. It moves from “Marginalization” and “Informing” to “Involvement,” “Collaboration,” and ultimately “Community Ownership” [cite: 65]. This shift requires public officials to pivot from being “gatekeepers” to “collaborators” and “process coordinators” [cite: 65].

Co-Creation Radar (CC-Radar)

The NordForsk COLDIGIT project developed the “Co-Creation Radar” to evaluate the quality of participatory processes [cite: 4, 66]. It was piloted in cities like Helsinki and Trondheim to help officials assess how citizen input is recognized and used in decision-making [cite: 66]. This focus on the “quality” of interaction ensures that digital tools do not just generate more data, but lead to better-quality, more equitable decisions [cite: 4].

Conclusions and the Path to Resilient Discourse

The technologies for enhancing democratic discourse and collective agency are maturing into a robust ecosystem of digital civic infrastructure. By aligning modular participation platforms with advanced deliberative tools and AI-driven synthesis, it is possible to scale deep, reason-giving dialogue to the level of thousands of participants [cite: 16, 32].

The key to a successful digital democracy is not the deployment of technology for the sake of efficiency, but the intentional design of tools built to enhance democratic engagement from connection to action [cite: 1, 40]. This requires:

• Decentralized Governance: Protecting user data and shifting power to municipal and citizen networks [cite: 6].
• Inclusive Design: Prioritizing safety and accessibility for vulnerable groups to counteract unequal representation [cite: 19].
• Traceability: Establishing clear, transparent loops between public input and institutional implementation [cite: 5, 17].
• Human Agency: Designing “scaffolds” that empower individuals and collectives rather than “crutches” that automate thought [cite: 8, 40, 50].

As AI becomes an integral part of people’s decision-making, the primary question is no longer how well it performs, but how it reshapes the conditions under which human agency operates [cite: 7, 47]. With strategic investment in “Digital Public Infrastructure” and a commitment to democratic values, these tools can restore trust in institutions and defend the fundamental rights and freedoms of the global citizenry [cite: 25, 40, 67].

---

1. A Framework for Digital Civic Infrastructure – Ash Center, https://ash.harvard.edu/resources/a-framework-for-digital-civic-infrastructure/
2. Digital democracy – European Parliament, https://www.europarl.europa.eu/RegData/etudes/BRIE/2020/646161/EPRS_BRI(2020)646161_EN.pdf
3. Decidim, a Technopolitical Network for Participatory Democracy – OAPEN Library, https://library.oapen.org/handle/20.500.12657/87634
4. Democratic innovation and digital participation: Harnessing collective intelligence for 21st-century decision-making – CHANSE, https://chanse.org/wp-content/uploads/Democratic_innovation_and_digital_participation.pdf
5. Decidim – Wikipedia, https://en.wikipedia.org/wiki/Decidim
6. Empowering Digital Democracy – Cambridge University Press & Assessment, https://www.cambridge.org/core/services/aop-cambridge-core/content/view/D39CBD8C8061EA6EE1B0FE0D64FA9E5D/S1537592724000409a.pdf/empowering-digital-democracy.pdf
7. Cultivating Human Agency | Prioritizing Autonomy – Imagining the Digital Future Center, https://imaginingthedigitalfuture.org/reports-and-publications/human-resilience-in-the-age-of-ai/cultivating-human-agency/
8. AI and the future of human decision-making – Deloitte, https://www.deloitte.com/us/en/insights/topics/talent/human-capital-trends/2026/decision-making-with-ai.html
9. AI multilingual tools for social innovation: new perspectives on civic technologies, digital empowerment and inclusive governance – Frontiers, https://www.frontiersin.org/journals/political-science/articles/10.3389/fpos.2026.1792788/full
10. Collective Intelligence Tools: The Complete Buyer’s Guide (2026) | Mindhive Blog, https://mindhive.ai/blog/collective-intelligence-tools
11. Fact-checking at a crossroads: Fact checkers’ perspectives on Community Notes, AI integration, and design recommendations | HKS Misinformation Review, https://misinforeview.hks.harvard.edu/article/fact-checking-at-a-crossroads-fact-checkers-perspectives-on-community-notes-ai-integration-and-design-recommendations/
12. 4 Useful Resources on Digital Participation Tools – Democracy Technologies, https://democracy-technologies.org/getting-started/resources-on-digital-participation-tools/
13. Here are the best digital participation tools of 2025 – People Powered, https://www.peoplepowered.org/news-content/best-digital-participation-tools-2025
14. Practical AI Use Cases for Community Managers – Higher Logic, https://www.higherlogic.com/blog/practical-ai-use-cases-for-community-managers/
15. AI in Public Deliberation: Technological Frontiers for Democracy Transition, https://www.frontiersin.org/research-topics/68574/ai-in-public-deliberation-technological-frontiers-for-democracy-transition
16. AI for Deliberation and Consensus – Stanford Digital Economy Lab, https://digitaleconomy.stanford.edu/project/ai-for-deliberation-and-consensus/
17. Alternative Digital Platforms and the Renewal of the Public Sphere: Decidim and the Democratic Governance of Participatory Infrastructures – MDPI, https://www.mdpi.com/2076-0760/15/3/166
18. Course: Designing Democratic Engagement with AI: Practical Strategies for Public Participation – InnovateUS, https://innovate-us.org/course/democratic-engagement-for-ai-era/
19. Inclusive AI-Enhanced Civic Engagement: Empowering Marginalized Voices – MDPI, https://www.mdpi.com/2075-4698/16/4/115
20. The Ecosystem of Deliberative Technologies for Public Input – Ash Center, https://ash.harvard.edu/resources/the-ecosystem-of-deliberative-technologies-for-public-input/
21. From Learning to Doing: An AI Coach for Public Engagement – Reboot Democracy, https://rebootdemocracy.ai/blog/ai-coach-for-public-engagement
22. Comparison of Decidim and Consul, https://decidim.org/blog/2019-01-14-consul-comparison/
23. Transforming Public Governance for Digital Democracy: The OECD Reinforcing Democracy Initiative | OECD, https://www.oecd.org/en/publications/the-oecd-reinforcing-democracy-initiative_9543bcfb-en/full-report/component-8.html
24. The best participatory democracy platforms worldwide – Blog Concorder, https://blog.concorder.net/en/the-best-participatory-democracy-platforms-worldwide/
25. Democracy – TICTeC – mySociety, https://tictec.mysociety.org/category/democracy/
26. Digitized Democracy: How AI is Reshaping Governance and Civic Life, https://www.nationalcivicleague.org/ncr-article/digitized-democracy-how-ai-is-reshaping-governance-and-civic-life/
27. vTaiwan, https://info.vtaiwan.tw/
28. vTaiwan’s hybrid approach to digital deliberation with AI – People Powered, https://www.peoplepowered.org/news-content/digital-participation-case-study-taiwan
29. vTaiwan – Crowdlaw for Congress, https://congress.crowd.law/case-vtaiwan.html
30. vTaiwan – Participedia, https://participedia.net/method/vtaiwan
31. Deliberative policymaking during COVID-19 | 9 | The case of Taiwan | H – Taylor & Francis eBooks, https://www.taylorfrancis.com/chapters/oa-edit/10.4324/9781003215929-9/deliberative-policymaking-covid-19-helen-liu-tze-luen-lin
32. Realizing the Potential Gains of AI-Enabled Deliberative Democracy, https://carnegieendowment.org/research/2026/05/realizing-the-potential-gains-of-ai-enabled-deliberative-democracy
33. Can AI Mediation Improve Democratic Deliberation? – arXiv, https://arxiv.org/html/2601.05904v1
34. Deliberation.io – Citizen Engagement Platform, https://deliberation.io/
35. Best AI Moderated Research Tools 2026: 14 Platforms Compared – Listen Labs, https://listenlabs.ai/articles/best-ai-moderated-research-tools/
36. A Guide to AI Moderation in UX and Product Research – Maze, https://maze.co/collections/ai/ai-moderation/
37. Using Civic Technology to Facilitate Meaningful Engagement – Commons Library, https://commonslibrary.org/civic-technology-for-meaningful-engagement/
38. AI’s Role in Deliberative Discussion | PS: Political Science & Politics | Cambridge Core, https://www.cambridge.org/core/journals/ps-political-science-and-politics/article/ais-role-in-deliberative-discussion/815CF6FE51CCBC5D0ACE918D249DD270
39. Artificial Intelligence and Improving Democratic Skills – BYU Social Science and AI Lab, https://sail.byu.edu/artificial-intelligence-and-improving-democratic-skills
40. Deliberative Muscles & AI – DemocracyNext, https://www.demnext.org/action/deliberative-muscles-and-ai
41. Democratic Roadmap: Building Civic Resilience to the Global Digital Information Manipulation Challenge – State Department Home, https://2021-2025.state.gov/roadmap-info-integrity/
42. AI and Democracy: Mapping the Intersections | Carnegie Endowment for International Peace, https://carnegieendowment.org/research/2026/01/ai-and-democracy-mapping-the-intersections
43. The trust-consensus paradox: why decentralized fact-checking faces challenges on polarizing topics – Institute for Strategic Dialogue, https://www.isdglobal.org/digital-dispatch/the-trust-consensus-paradox-why-decentralized-fact-checking-faces-challenges-on-polarizing-topics/
44. Cofacts – Rights CoLab, https://rightscolab.org/case_study/cofacts/
45. Full article: Mapping Global Emerging Scholarly Research and Practices of AI-supported Fact-Checking Tools in Journalism – Taylor & Francis, https://www.tandfonline.com/doi/full/10.1080/17512786.2025.2463470
46. Civic Tech – commercial innovation opportunities to reduce misinformation: research report, https://www.gov.scot/publications/commercial-innovation-opportunities-civic-tech-reduce-misinformation-thinking-outside-bunk/
47. The Future of Human Agency | Pew Research Center, https://www.pewresearch.org/internet/2023/02/24/the-future-of-human-agency/
48. Artificial Intelligence and Agency: Tie-breaking in AI Decision-Making – PMC – NIH, https://pmc.ncbi.nlm.nih.gov/articles/PMC10980648/
49. Full article: Learner agency in self-directed use of GenAI for informal language learning, https://www.tandfonline.com/doi/full/10.1080/17501229.2026.2648781
50. Redefining Agency: A Capability-Driven Research Agenda for Generative AI in Education, https://www.mdpi.com/2227-7102/16/1/155
51. From Efficiency to Deliberation: Rethinking AI’s Role in Institutionalizing Democratic Innovations – Cogitatio Press, https://www.cogitatiopress.com/politicsandgovernance/article/download/10632/4752
52. Reducing political polarization through conversations with artificial intelligence | Journal of Computer-Mediated Communication | Oxford Academic, https://academic.oup.com/jcmc/article/31/2/zmag003/8551324
53. Using AI persuasion to reduce political polarization – EconStor, https://www.econstor.eu/bitstream/10419/334534/1/1947924397.pdf
54. Leveraging AI for democratic discourse: Chat interventions can improve online political conversations at scale – PMC, https://pmc.ncbi.nlm.nih.gov/articles/PMC10576030/
55. Leveraging AI for democratic discourse: Chat interventions can improve online political conversations at scale | PNAS, https://www.pnas.org/doi/10.1073/pnas.2311627120
56. Using AI persuasion to reduce political polarization – IDEAS/RePEc, https://ideas.repec.org/p/zbw/zewdip/334534.html
57. Data Sources & Methodology – Digital Democracy | CalMatters, https://calmatters.digitaldemocracy.org/data-sources-methodology
58. A growing observatory of examples of how open data from official sources and generative artificial intelligence (AI) are intersecting across domains and geographies. – Policy Repository, https://repository.opendatapolicylab.org/genai/
59. GitHub – npstorey/civic-ai-tools: Open-source platform connecting AI assistants to government open data — MCP server, curated civic MCP directory, and anti-hallucination framework for all 559 Socrata portals, https://github.com/npstorey/civic-ai-tools
60. The futures of digital democracy – PMC – NIH, https://pmc.ncbi.nlm.nih.gov/articles/PMC13087570/
61. (PDF) Digital Democracy: Enhancing Democratic Participation and Political Transparency Through Digital Technologies – ResearchGate, https://www.researchgate.net/publication/383485405_Digital_Democracy_Enhancing_Democratic_Participation_and_Political_Transparency_Through_Digital_Technologies
62. Impact of Information and Communication Technologies on Democratic Processes and Citizen Participation – MDPI, https://www.mdpi.com/2075-4698/15/2/40
63. UN E-Government Survey 2024 – the United Nations, https://publicadministration.un.org/egovkb
64. E-Government Survey 2024 – DESA Publications, https://desapublications.un.org/sites/default/files/publications/2024-09/(Web version) E-Government Survey 2024 1392024.pdf
65. Tools for Participatory Democracy: Spectrum of Community Engagement to Ownership, https://www.participatorybudgeting.org/tools-for-participatory-democracy-spectrum-of-community-engagement-to-ownership/
66. Collective Intelligence through Digital Tools (COLDIGIT) – NordForsk, https://www.nordforsk.org/node/1655
67. Digital democracy in action: 5 lessons to support resilient democracies, https://www.wfd.org/commentary/digital-democracy-action-5-lessons-support-resilient-democracies