AI as Political Intermediary
Large language models (LLMs) have rapidly become everyday interfaces for political questions. By mid-2025, ChatGPT alone had roughly 700 million weekly users exchanging 18 billion messages per week, with non-work uses accounting for more than 70% of consumer activity. Citizens now routinely ask AI models to explain ballot measures, summarize court rulings, draft messages to elected officials, translate bureaucratic language, and make sense of fast-moving public events. Yet these exchanges occupy an unusual place in political communication research. Surveys capture what people report doing; social media traces capture what people choose to say in public. LLM prompts capture something different: private or semi-private, task-oriented requests for help with political information, civic action, and public institutions.
This study analyzes 4.30 million human–AI conversations from three large public datasets—WildChat, LMSYS-Chat, and ShareChat—spanning 2023 to 2025. Two validated LLM classifiers identify political content, use case, and expressed ideology in user messages. The central finding: political content appears in 3.9% of conversations, varies sharply by platform publicness and conversation depth, and is overwhelmingly practical. Users ask for information, draft text, and process documents far more often than they state opinions. A regression-discontinuity-in-time design around the 2024 U.S. presidential election result call shows that the call changed the expressive subset: among U.S. users, stance-taking, affective language, and ideological extremity rose; comparable conversations elsewhere did not. AI conversation is less a public square than a conversational political intermediary, absorbing routine demand and becoming expressive when major events make political stakes explicit.
The Intermediary View: Beyond Public Expression
Most research on AI and politics has treated models as objects of evaluation or sources of influence—examining their ideological tendencies, their ability to simulate public opinion, or whether AI-generated messages can change attitudes. This study shifts focus to the initiating side of the interaction: the prompt itself is a political communication trace that reveals the issue, task, audience, language, and institutional problem a user chooses to bring to an AI system.
If LLMs are becoming venues for expressive politics, political use should resemble the public-facing dynamics of digital platforms: opinion, argument, identity signaling, persuasion, and mobilization. If they are instead becoming civic intermediaries, political use should look more practical: explanation, summarization, translation, drafting, and help navigating institutions. Both uses can coexist, but their balance indicates whether LLMs extend the expressive politics of platform publics or absorb the quieter informational and administrative burdens of everyday citizenship.
The evidence strongly supports the intermediary account. Across 248,935 political user turns with use-case labels, 64.7% seek information or explanation, 13.0% involve writing or drafting, and 11.8% involve document processing. Opinion and argument account for just 9.4%. Mobilization, party organizing, and explicit public-agency or legal help together account for just over 1%. The modal political exchange with AI is not public voice but assistance.
This practical skew is not an artifact of the broad political definition used. Even after removing the most administrative categories—government-services topics, document-processing turns, and civic or legal help—information seeking and writing or editing still account for 88.3% of political turns, while expressive use rises only to 11.6%.
Where Political AI Use Appears
Political use is a recurring but minority share of activity in all three corpora: 3.3% of WildChat conversations, 3.9% of LMSYS-Chat conversations, and 18.7% of ShareChat conversations. The five- to six-fold gap across platforms reflects the selection mechanisms that make conversations publicly observable. ShareChat consists of conversations users chose to make publicly shareable, so it carries more opinion and identity-laden material. WildChat and LMSYS-Chat capture broader service and evaluation settings where many ordinary tasks never acquire an audience.
Political prevalence also rises monotonically with conversation depth. In WildChat, it increases from 2.6% of single-turn exchanges to 12.4% of conversations with eleven or more user turns. Political conversations are longer on average than non-political ones, placing political AI use closer to problem work than to one-shot lookup. A simple factual query can often be resolved in one turn; understanding a policy, revising an appeal, translating an official notice, or deciding how to frame a complaint often requires back-and-forth clarification.
Geographically, WildChat's metadata reveals that political share varies substantially across world regions and is higher in several lower-income and more politically constrained settings than in wealthy democracies. Sub-Saharan Africa has the highest regional prevalence at 5.8%, followed by Latin America and the Caribbean at 4.7% and Eastern Europe and Central Asia at 4.0%. North America is lower at 2.3%. Political share rises from 3.1% in democratic or free contexts to 4.0% in authoritarian or not-free contexts, and from 2.9% in open-internet contexts to 3.9% in closed-internet contexts. Information seeking also becomes more common in more constrained contexts. These descriptive patterns suggest AI chat may matter most politically where it is not merely another convenience technology, but a partial substitute for more costly or constrained channels of political information and bureaucratic navigation.
The Expressive Margin: When Events Activate Political Voice
The descriptive contrast points to a sharper implication. If most political use is practical, then ordinary AI conversations should not always behave like public-facing political expression. But if expressive use is where partisanship and affect concentrate, then major political events should leave their clearest mark on that subset of interaction.
The 2024 U.S. presidential election provides a test. The election date was known in advance, but the winner was not publicly settled until the Associated Press called the race for Donald Trump on November 6, 2024. Using the AP call as the cutoff in a regression-discontinuity-in-time design, U.S. conversations immediately before and after the call were compared, with conversations outside the U.S. serving as a geographic comparison series.
The result did not substantially increase how often Americans used AI for politics. It changed the form of that use. After the AP call, U.S. political conversations became more likely to express an opinion (an 8.5 percentage-point increase), more affectively charged (a 6.4 point increase), and more ideologically extreme. No comparable shift appears outside the United States. A single event thus pushed one-to-one AI conversation toward the partisan and affective expression that political events are known to activate in surveys and public platforms.
Among U.S. users who expressed a stance, affective polarization rose by 0.55 scale units on a 0–3 affect scale—a substantively large shift. This indicates that the result changed not only whether users expressed positions, but also how they framed political disagreement. The finding is consistent with work showing that partisan conflict often operates through social identity, threat, and dislike rather than through policy distance alone.
Crucially, the U.S. expressive effect was not confined to election-topic prompts. Among political conversations that were not about the election, stance-taking still increased by 8.5 points—nearly the full headline effect. For U.S. users, the result call mattered because it made the winner-loser cue explicit, and that cue traveled into policy, courts, immigration, foreign affairs, party conflict, and ideological identity.
Theoretical Contributions
The paper makes two main contributions. Methodologically, it develops and validates a procedure for measuring political content, topic, purpose, and ideological expression in large multilingual corpora of human–AI conversation. These traces reveal a form of political communication that surveys and social-media data largely miss. At the same time, they should not be treated as off-the-shelf attitude data: most political prompts do not express a position, and the prompts that do are selected by platform, task, and event context.
Theoretically, the paper locates LLMs as practical intermediaries in political communication. Citizens use them to understand, produce, and act on politics, and these intermediaries follow two regularities familiar from older venues: political attention sorts unevenly across them, and public events alter the character of political expression within them. The point is not simply that politics appears in AI conversation. It is that a private, task-oriented, conversational venue absorbs routine civic demand while also registering the partisan and affective shocks of public events.
The intermediary view connects LLMs to three strands of political communication scholarship: everyday political talk, which shows citizens reason about politics through ordinary conversation; research on changing information environments, which argues political communication now moves through fragmented, high-choice systems; and the study of administrative burden, which shows routine dealings with public institutions can be costly, confusing, procedural, and textual. Human–AI conversations make this kind of political behavior more visible: a substantial share of political AI use consists of document-centered and institution-facing work that these theories emphasize but existing data rarely capture directly.
Conclusion
Far from being either a new public square or a mere information tool, LLMs function as dual-purpose political intermediaries. Most of the time, they absorb routine civic demand—explaining policies, drafting documents, translating official language, and helping users navigate institutions. But when a major political event makes winners, losers, and stakes explicit, the same interface can become a venue for partisan expression, affective charge, and ideological positioning. Understanding this dual role is essential for scholars, platform designers, and citizens alike as AI becomes an increasingly common interface between people and public life.
