Artificial intelligence is becoming a provocative new tool in political campaigns, enabling parties to build synthetic focus groups composed of AI-generated voter personas. As midterm elections approach in the United States and other countries, researchers are raising critical questions about the ethical and practical implications of relying on such synthetic voters, particularly around the authenticity of public opinion and the risks to democratic discourse.
What Happened
A recent pre-print study titled “Fake Plastic Voters: When Can Political Parties Use AI Simulated Focus Groups?” published in 2025 by Claudio Novelli and his team at Yale’s Digital Ethics Lab examined the use of AI-generated synthetic voters in campaign research. Their work proposes a decision matrix to guide political parties on when the deployment of AI-simulated focus groups is appropriate and when it is too risky. The research highlights limitations and provides cautionary frameworks for campaigns considering AI-based population simulations as part of their strategy.
Key Facts
The study emphasizes two primary use cases for focus groups: understanding voter perspectives and testing campaign messages. It advises against using AI-simulated groups to gauge how voters truly feel or make sense of political realities due to a lack of nuance and authentic human interaction, including nonverbal cues and emotional responses. However, it leaves room for AI use in lower-risk scenarios such as internal message testing, provided the synthetic populations statistically approximate targeted voter groups rather than attempting to replicate individuals.
The study distinguishes between different qualities of synthetic voters—ranging from generic personas generated by large language models (LLMs) to highly detailed digital twins of real individuals—and warns of legal and computational risks with the latter. The most practical and ethical application falls in the middle: using AI to simulate diverse, statistically representative populations without targeting or replicating actual persons.
Additionally, Jennifer Stromer-Galley, a professor of information studies at Syracuse University, noted that both traditional and digital focus groups have historically struggled to truly represent public opinion, often excluding marginalized voices and over-representing specific demographics. Current AI training data, drawn from platforms like Twitter and Reddit, is criticized for reflecting extreme or average viewpoints but lacking comprehensiveness, which risks further distorting political messaging and the public’s perception.
What This Means
The integration of AI-simulated voters in political campaigns introduces significant challenges for both political strategists and democratic societies. While AI-generated synthetic personas could enhance the efficiency of message testing and outreach by modeling diverse demographics, they pose serious risks if over-relied upon for gauging genuine public sentiment. The subtleties of human communication—body language, spontaneous reactions, and nuanced social cues—are absent in synthetic simulations, potentially leading campaigns to misinterpret or oversimplify voter priorities.
Moreover, the reliance on synthetic data risks reinforcing a feedback loop wherein political elites “create” public opinion through artificial reflections rather than responding to authentic voices. As AI tools shape campaign narratives, this could erode the shared reality necessary for healthy democratic debate, potentially skewing policy discussions and voter expectations based on artificial constructs. Voters and policymakers alike should be aware of these evolving dynamics as synthetic data increasingly influences political decision-making.
Background
Focus groups have long been a staple of political campaigns, aiding candidates in message crafting and understanding voter clusters. However, the inherent limitations of traditional focus groups—such as sampling bias and artificial settings—have been well documented. The advent of AI-generated synthetic voters adds a new dimension, promising scalability and diversity but also raising novel ethical and representational concerns as highlighted by scholars like Novelli and Stromer-Galley.
What Remains Unclear
Key uncertainties include how regulations or campaign standards will evolve to govern the use of synthetic voters, especially digital twins, given their legal and privacy implications. The extent to which campaigns will adopt these tools widely in upcoming elections, such as the 2028 U.S. presidential race, remains to be seen. Furthermore, it is unclear how these synthetic focus groups will be scrutinized for accuracy or bias and whether the public will have recourse or transparency about their use in shaping political messaging.
What Comes Next
Experts predict experimental adoption of AI-simulated voters in key Senate races in the current midterm cycle, with broader integration expected by the 2028 U.S. presidential election. Ongoing research like Yale’s decision matrix aims to inform campaign strategies with guidelines to mitigate risks. Monitoring how these technologies develop and are regulated will be crucial in the coming years as AI tools reshape political communications.
Sources
This article is based on reporting and publicly available information from the following sources:
Read more AI Regulation stories on Goka World News.