By Aisha Nayla
The emergence of artificial intelligence and its power go hand in hand with growing public anxiety about job security, media literacy, and almost every sector of everyday life. According to a 2024 Pew Research Center survey, 39 percent of Americans were highly concerned that AI would be mostly used for bad purposes during the 2024 presidential campaign, with only 5 percent believing it would be used for good purposes. Current public consensus reflects growing fears over AI’s role in an increasingly digitized democracy, and for good reason. The increased use of AI-generated content and the shift of several countries including the Philippines, Myanmar, and the United States towards digital authoritarianism by embracing tools such as AI-powered mass surveillance raises the stakes more than ever.
Beyond the complexities posed to the information climate, the ability for faster content creation also holds the potential to impact democratic discourse and challenge the integrity of elections. But while these concerns have merit, what’s often missing from the conversation is AI’s potential to strengthen democratic systems and practices; how local governments, consulting platforms, and civic societies are using it as a means of democratic innovation, whether it be in the election space, the integration of digital governance, and more.
I conducted this research as part of my internship at the National Civic League’s Center for Democratic Innovation, which looks into understanding new and improved ways to make democracy more participatory, equitable, and productive. As a political science major who also studies communication technology platforms and practices, I have always been intrigued by the ways AI has penetrated and shaped the civic space, as well as in exploring its potential to improve it. And yet, the realm of AI itself is inherently broad, and it can be difficult to grapple how it interacts with our democratic systems without a closer inspection of specific areas and case studies.
The purpose of this article is to help readers better understand the status quo of how AI is being used in government and civic spaces and to shed light on the challenges and opportunities it may pose for the future of an increasingly digitized democracy in the United States and beyond. My research process consisted of four primary stages.
- Developing an annotated bibliography of credible sources that discussed AI in governance, the challenges of digital democracy, and its current influence in elections and government decision-making.
- Identifying more specific case studies that spotlighted how AI was being integrated into various governments worldwide, which provided a microlens on its implementation and impact in varying regional contexts.
- Extracting significant data points from these initial stages to quantify trends, challenges, and successes.
- Synthesizing these findings into a summary report of key takeaways, which highlighted emerging themes, gaps, and policy implications and ultimately led to the development of this article.
Three Areas of AI in Democracy
My research identified three primary areas in which AI is being increasingly used in democratic life:
- Digital governance and public services
- Election outreach
- Civic engagement
After conducting an extensive literature review, these three areas consistently demonstrated to be key intersections of AI in American democracy; they reflect both the operational and participatory functions of a democracy, namely how governments function, how they connect with their citizens, and in turn, how citizens engage with and shape policy. Whether it be managing infrastructure and administering public services, or organizing elections and facilitating a public input system, each area demonstrates a unique yet interconnected facet of democratic governance. Together, they provide a holistic view of AI’s potential to leverage or disrupt transparency, equity, and civic empowerment, core values that lie in any functioning democracy.
Digital Governance and Public Services
Many local governments have started moving towards digital governance, using AI to better inform and streamline administrative services. In Dearborn, Michigan, where more than half the residents speak a language other than English at home, AI translation tools are currently being explored to improve the city website’s accessibility. On the website, virtual AI agents are delivering personalized information and interactions in Arabic, English, and Spanish, while websites and documents have been translated to further expand the delivery of crucial government information and services.
Phoenix, Arizona, has developed “myPHX311,” a website and app that answers common questions in Spanish and English and connects people to local agencies that can best resolve their issues. It has become an efficient way to request public records, start or stop water service, and report incidents i.e. streetlight outages, property violations, and more. This translates into the day-to-day operations that keep the area running efficiently.
Lastly, in Washington, D.C., city workers are using an AI system to analyze videos that help inspect 1,800 miles of sewer pipes. While manual reporting used to take 75 minutes, these automated inspections only take 10 minutes.
Outside the United States, AI is also being adopted for urban planning. In Shanghai and New Zealand, AI-driven “digital twins,” a digital representation of a physical asset or system, enables city planners to better visualize development projects and predict their impacts, such as how they would affect surrounding neighborhoods. In the Philippines, local officials have used AI to analyze news and social media content in Tagalog and English, which help them better understand public development priorities.
Voter Outreach
When it comes to campaigning and voter outreach, the technology has been adopted as a means of expanding outreach and more strategic targeting. In New York City, Mayor Eric Adams attempted to reach the city’s non-English-speaking communities through AI-powered robocalls. The technology presents Adams’ voice in various languages he does not actually speak, such as Mandarin and Yiddish, to promote city hiring events. The watchdog group Surveillance Oversight Project criticized these robocalls as an unethical use of AI that was misleading. Adams defended its use by saying “I got one thing: I’ve got to run the city, and I have to be able to speak to people in the languages that they understand, and I’m happy to do so.” The ethics of deepfakes continue to be a highly contentious area, especially when used in politics.
Perhaps the most significant use of AI in elections is within the political consulting industry. Companies such as Votivate are promoting a future of campaign tech that leverages real-time voter data for use in advanced modeling and analytics; through AI algorithms, candidates are able to create more personalized and targeted advertisements, tailored to the interests and behaviors of individual voters. Additionally, AI-driven social media analysis allows campaigns to better monitor public sentiment, identify effective influencers, and craft the right narratives to sway public opinion. Some other tools offered by the platform include data on unregistered voters, managing 20,000+ doors on a map to ease door-to-door canvassing, and providing intuitive data visualizations.
Civic Engagement
And yet, these platforms aren’t merely benefiting candidates. Go Vocal, for instance, offers a platform designed for data-driven decision making and more inclusive governance. They collect resident input through surveys and consultations, then leverage large language models (LLMs) to provide advanced analytics and track public sentiment. This helps local governments grasp how opinion shifts from neighborhood to neighborhood, so they can better reach residents where they are. For instance, St Louis was able to involve 16,000+ residents in the allocation of a settlement fund, through a digital community engagement platform developed by Go Vocal.
In Allen, Texas, a hybrid engagement strategy successfully engaged 612 households and received a nearly representative sample in the process of its “smart city” development for 2045 through tools such as automated communications with residents. Another platform, Pol.is, tracks public opinion by functioning as an online forum where users can post comments on a topic, but are limited to upvoting or downvoting instead of replying. This data is then used to generate a map of participants in the debate that clusters together those who have voted similarly – demonstrating where there is divide and consensus. As of August 2018, the website claims it had been used in 26 cases, with 80 percent leading to “decisive government action.” For instance, it inspired regulations for Uber, which had received fierce opposition from local taxi drivers, and led to an act that created a space for small-scale start-ups in Taiwan.
Opportunities for Innovation in Governance
While the impacts of AI’s usage in these sectors are still being assessed, especially as technology continues to grow and change at such a rapid rate, there seems to be a clear trend of what opportunities it brings to our democratic functions: efficiency, accessibility, and engagement. Whether it be Washington D.C.’s sewer pipe inspections or Texas’ automated communications with residents, operational efficiency has been a key outcome of several AI-driven technologies that are being implemented by local and national governments. The time it takes to analyze videos to inspect sewer pipes and send payment reminders has been significantly reduced, paving the way for more innovation. But this efficiency extends beyond cutting time and service delivery; it also means better reaching the communities that need to be reached—whether that’s for voter targeting, communicating in a language understood by diverse populations, or other functions. Top-down communication is better enhanced, which also means better bottom-up accessibility. For instance, the emergence of several translation tools used in Arizona’s “myPHX311” platform and Dearborn’s city website means improved outreach to Spanish and Arab-speaking populations, thus improving these groups’ access to government services such as payment portals, water services, reporting violations, and more.
Taken together, these outcomes empower civic engagement. With public participation platforms such as Go Vocal and Polis, the leveraging of large language models to better process public opinion has allowed governments to better understand what their citizens need and how to serve them, which lies at the core of a functioning democracy. This enhanced understanding may lead to more effective investing in the most relevant sectors of public life, and through developments such as AI translation tools, such resources and services can be appropriately used.
Challenges: Bias, Deepfakes, and More
As with all new and emerging technologies, especially one that is ever-changing, the use of AI in democratic innovation is not without challenges. Given that much of AI works around existing datasets, there are fears that AI algorithms may unintentionally favor certain demographics when it comes to providing recommendations for policy and understanding public opinion, which can become dangerous if a large portion of the decision-making process is influenced by it. When it comes to this issue, the need to train AI with more diverse data sets becomes ever more important, and that can only occur with further research. Beyond creating false content, generative AI such as deepfakes also erodes trust in real media, enabling public figures to dismiss authentic recordings as fake and avoid accountability.
With deepfakes becoming increasingly realistic and people becoming more aware of that, false claims of real content being AI-generated are becoming more common. While Mayor Eric Adams argued that his robocalls were a necessary measure to reach a greater range of the population, the skepticism around using generative AI for civic engagement has validity. Lastly, while all 50 states have introduced AI-related legislation, the lack of federal standards may lead states to approach AI regulation without consistency, which may result in potential gaps in oversight and accountability.
Conclusion
Clearly, AI has the potential to transform how our democracy functions, but there are important questions that remain unanswered. Going forward, it is worthwhile not only to acknowledge what has been accomplished in the realm of democratic innovation, but also critical to address an array of pressing questions: While platforms like Go Vocal and Pol.is can efficiently summarize public opinion, what does this mean for deeper dialogue that warrants more nuance? Do these platforms risk reducing complex sentiment into static data points? Further, given the potential for dire consequences, what safeguards are most effective against algorithmic bias in frontier AI systems, and how can they be enforced when using AI for high-stakes decision making?
Lastly, how might AI tools be co-designed with communities to truly reflect local priorities? Currently, most of the AI infrastructure is developed through a top-down approach by governments or private firms, but it is crucial to further explore how communities themselves can be co-developers in these algorithms and criteria used for civic technologies. While not an exhaustive list, a closer inspection into these remaining questions can pave the way to truly integrate AI as an empowering tool for civic engagement and in facilitating a democracy that functions not just in theory, but in practice.
Aisha Nayla (she/her) is a student from Jakarta, Indonesia, studying Political Science and Communications at the University of Southern California. Her primary interests lie in the intersection of technology and democracy, as well as religion and governance, especially in Southeast Asia.
