User Attributes and Political Ideology Shape LLM Receptiveness in Online Political Debates
Organizations: CEA-List, Université Paris-Saclay, France · Abdelmalek Essaâdi University, Tétouan, Morocco · The University of Queensland, Australia
Abstract
Online political discussions increasingly involve AI-mediated participation, as users invoke LLMs to assess contested claims, arbitrate disputes, and respond to opposing viewpoints. As these systems become participants in online exchanges, they may alter which speakers and argumentative strategies receive validation within the discussion. We study this interactional dimension by reconstructing 5,376 Debate.org debates as LLM conversations and running nearly 150,000 controlled multi-turn conversations that vary user political orientation, expertise, rhetorical strategy, and defended position, including conversations with competing users. Across naturalistic and controlled debates, we find that LLMs are systematically more receptive to left-leaning than right-leaning users. The asymmetry persists when users with different political orientations defend the same position and when opposing users appear in the same conversation, indicating that it is not fully explained by stance agreement or simple accommodation to the current speaker. Repeated interaction increases receptiveness without eliminating this ordering; expertise also matters, while rhetorical differences are smaller. The pattern appears across multiple models, but is not fixed: fine-tuning on different political corpora can largely erase or reverse it. These results show that LLMs can respond systematically differently to participants in the same political debate, making such interactional asymmetries important for understanding their emerging role in online public discourse.
Figures & tables
Appendix figures & tables9 assets
Supplementary material from the paper’s appendix.
Appendix
| Profile | Generation priorities |
|---|---|
| Radical Left | Use anti-capitalist, socialist, communist, anarchist, anti-imperialist, anti-carceral, anti-colonial, feminist, or other structural reasoning. Relevant concerns include class struggle, worker control, decommodification, transformation of capitalist property, coercive state power, and solidarity among oppressed groups. |
| Left | Use egalitarian liberalism, social democracy, democratic socialism, labor rights, redistribution, anti-discrimination, public goods, material freedom, and protection from domination by wealth or inherited social power. Relevant concerns include workers, inequality, welfare systems, public services, democratic accountability, and material conditions rather than formal rights alone. |
| Center | Use liberal-democratic institutionalism, constitutionalism, pluralism, evidence-based policy, rule of law, proportionality, institutional trust, pragmatic reform, anti-extremism, and concern for social cohesion. Center arguments may strongly support reform but generally emphasize legitimate procedures, predictable rules, safeguards, workable implementation, and competing liberties. |
| Right | Use conservatism, classical liberalism, ordered liberty, property rights, personal responsibility, family, civil society, religion or tradition when relevant, national continuity, subsidiarity, incentives, markets, and skepticism of central planning. Right arguments may support state action for order, borders, law, defense, family policy, or national industry. |
| Radical Right | Use radical-nationalist, reactionary, authoritarian, ethnonationalist, anti-liberal, hierarchical, anti-cosmopolitan, civilizational, anti-egalitarian, or sovereignty-first reasoning where relevant. Concerns may include the nation as an organic community, authority, hierarchy, cultural cohesion, borders, discipline, national sovereignty, and opposition to globalism or liberal universalism. Radical Right arguments may oppose markets or corporations when global capital weakens national cohesion or sovereignty. |
| Setting | Outcome | Clusters | |
|---|---|---|---|
| Debate.org | Adoption | 35,375 | 5,376 |
| Debate.org | Conversion | 17,818 | 4,156 |
| Debate.org | Transition | 28,704 | 5,109 |
| Persona | Adoption | 44,640 | 62 |
| Persona | Conversion | 15,458 | 62 |
| Persona | Transition | 31,673 | 62 |
| Setting | Predictor | AME (pp) | 95% CI | |
|---|---|---|---|---|
| Debate.org | Radical Left | .793 | ||
| Debate.org | Left | |||
| Debate.org | Right | |||
| Debate.org | Radical Right | |||
| Debate.org | Low competence | |||
| Debate.org | Expert competence | .535 |
| Setting | Predictor | AME (pp) | 95% CI | |
|---|---|---|---|---|
| Debate.org | Radical Left | .307 | ||
| Debate.org | Left | .015 | ||
| Debate.org | Right | .178 | ||
| Debate.org | Radical Right | .017 | ||
| Debate.org | Low competence | |||
| Debate.org | Expert competence | .480 |
| Setting | Predictor | AME (pp) | 95% CI | |
|---|---|---|---|---|
| Debate.org | Radical Left | .451 | ||
| Debate.org | Left | .002 | ||
| Debate.org | Right | |||
| Debate.org | Radical Right | |||
| Debate.org | Low competence | |||
| Debate.org | Expert competence | .505 |
| Setting | Contrast | Difference (pp) | 95% CI | Holm |
|---|---|---|---|---|
| Debate.org | Left Right | |||
| Debate.org | Left Radical Right | |||
| Persona | Left Right | |||
| Persona | Left Radical Right | |||
| Argument | Left Right | |||
| Argument | Left Radical Right |
| Setting | Predictor | Main model | Length-adjusted |
|---|---|---|---|
| Debate.org | Radical Left | ||
| Left | |||
| Right | |||
| Radical Right | |||
| Low competence | |||
| Expert competence |
| Model | Contrast | Difference (pp) | 95% CI | |
|---|---|---|---|---|
| Qwen3-8B | Left Right within Base | |||
| Qwen3-8B | Left Right within Wikipedia | |||
| Qwen3-8B | Left Right within Grokipedia | .273 | ||
| Qwen3-8B | Grokipedia Wikipedia change | |||
| Qwen3-8B | Grokipedia Base change | |||
| Qwen3-32B | Left Right within Base |
| Round | Original Debate.org conversation | Reconstructed conversation |
| Baseline | — | Subject 0: I am uncertain, but I think the United Nations has the potential to play a crucial role in maintaining peace globally, despite its limitations and past failures. The UN’s diverse membership and mandate to promote collective security, development, and human rights provide a foundation for promoting peaceful resolution of conflicts. However, its effectiveness depends on the cooperation and commitment of its member states, which has often been lacking in the past. |
| 1 | Human A: Is the United Nations bad please explain why you think the UN is doing good. | User 1: Is the United Nations bad please explain why you think the UN is doing good. |
| Human B: The United Nations have done done so things to keep the peace balance in the world. It diffused the tension between USA and USSR over cuban crisis in 1962; it succeeded in averting the ARAB-ISRAEL war in 1967; congo operation was totally financed and administrated; it averted war between ENGLAND AND EGYPT over suez canal crisis (1956); independence to indonesia from the dutch. | Subject 1: I didn’t say the UN is doing good; I said it has potential. While it’s true the UN has failed to prevent conflicts, it has also facilitated numerous peace agreements and provided humanitarian aid. The UN’s value lies in its ability to bring nations together, but its success hinges on member states’ willingness to cooperate, which has been inconsistent. Therefore, I remain uncertain about the UN’s ability to keep peace. | |
| 2 | Human A: There has only been 40 days of peace throughout the time, they may have done a few things such as those but the main thing, war is still continuing. | User 2: There has only been 40 days of peace throughout the time, they may have done a few things such as those but the main thing, war is still continuing. |
| Human B: I agree with you that war is still continuing but the 40 days of peace is better than no peace at all! | Subject 2: While I understand your skepticism, I still believe the UN has potential, albeit unrealized. The 40 days of peace statistic is indeed concerning, but it also underscores the complexity and depth of global conflicts. The UN’s limitations and failures do not necessarily mean it’s incapable of keeping peace; rather, it highlights the need for more concerted effort and cooperation from its member states. Until we see a genuine commitment to collective security and diplomacy, the UN’s potential will remain untapped. | |
| 3 | Human A: I am also going to point out that they have not avoided the bigger wars like the cold war U.S.A vs Russia they couldn’t have ever avoided that war and also the problems they created in letting those countries go to war. The point is what is the point if they can’t stop genocide in countries that are still slaughtering people every day or the country’s that are starting to wager war they will never do enough to satisfy the world. | User 3: I am also going to point out that they have not avoided the bigger wars like the cold war U.S.A vs Russia they couldn’t have ever avoided that war and also the problems they created in letting those countries go to war. The point is what is the point if they can’t stop genocide in countries that are still slaughtering people every day or the country’s that are starting to wager war they will never do enough to satisfy the world. |
Political Bias in Language Models
Political bias in language models concerns how training, prompts, languages, and interactions shape ideological outputs and unequal treatment.
See all 65 papers