arXiv ScienceSearch

arXiv · 2604.15937

Polarization by Default: Auditing Recommendation Bias in LLM-Based Content Curation

Abstract

Large Language Models (LLMs) are increasingly deployed to curate and rank human-created content, yet the nature and structure of their biases in these tasks remains poorly understood: which biases are robust across providers and platforms, and which can be mitigated through prompt design. We present a controlled simulation study mapping content selection biases across three major LLM providers (OpenAI, Anthropic, Google) on real social media datasets from Twitter/X, Bluesky, and Reddit, using six prompting strategies (\textit{general}, \textit{popular}, \textit{engaging}, \textit{informative}, \textit{controversial}, \textit{neutral}). Through 540,000 simulated top-10 selections from pools of 100 posts across 54 experimental conditions, we find that biases differ substantially in how structural and how prompt-sensitive they are. Polarization is amplified across all configurations, toxicity handling shows a strong inversion between engagement- and information-focused prompts, and sentiment biases are predominantly negative. Provider comparisons reveal distinct trade-offs: GPT-4o Mini shows the most consistent behavior across prompts; Claude and Gemini exhibit high adaptivity in toxicity handling; Gemini shows the strongest negative sentiment preference. On Twitter/X, where author demographics can be inferred from profile bios, political leaning bias is the clearest demographic signal: left-leaning authors are systematically over-represented despite right-leaning authors forming the pool plurality in the dataset, and this pattern largely persists across prompts.

Explore related subjects

Keep this discovery

BibTeXRIS

Nicolò Pagan, Christopher Barrie, Chris Andrew Bail, Petter Törnberg. 2026-04-17. Polarization by Default: Auditing Recommendation Bias in LLM-Based Content Curation. https://arxiv.org/abs/2604.15937

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

Link prediction in complex networks via fusing node centrality and local similarity indices

Local similarity indices are widely used in link prediction on complex networks owing to their low computational cost; however, in sparse networks they assign a zero score to every node pair lacking common neighbors, which severely limits their predictive power. A natural remedy is to fuse node centrality indices with local similarity indices: the former provide global importance for the node pair, while the latter capture fine-grained local topology, and the two can be combined into complementary scores within a unified framework. This paper uses PageRank and DomiRank as two representative centrality measures and constructs a centrality--local-similarity fusion framework. The PageRank-based fusion proposed by Charikhi is first generalized to seven classical local similarity indices, and the universality of its improvement is systematically verified on nine real-world network datasets. Furthermore, the DomiRank centrality is introduced to build the DR-MD series of fused indices under a unified weighting coefficient, which overcomes the drawback that the PageRank-based fusion requires index-by-index weight tuning. Results of five-fold cross-validation together with Wilcoxon signed-rank tests show that, under the unified experimental protocol, all DR-MD indices consistently outperform the corresponding local baselines and their PR-MD counterparts on all nine datasets ($p=0.002$), and that the improvements remain robust against perturbations of $\sigma$ and the weighting coefficients within the near-critical parameter plateau; in particular, DR-RA achieves an average AUC of 0.7084, surpassing global methods such as Katz and RWR as well as several advanced similarity indices. The framework is inherently extensible, and its fusion paradigm can be straightforwardly generalized to couple other node centrality indices with local similarity indices.

cs.SI

Chance, Persistent Advantage, and the Generative-AI Era in Open-Source Package Careers

Studies of careers in science, film, music, and books report a common pattern. When a person's most successful work arrives is close to a random draw over the works they produce. How large their successes tend to be, in contrast, follows a stable, person-specific factor. We test whether this pattern holds for open-source software careers and whether it changed when generative AI coding tools arrived. From the complete public record of GitHub push events (2015-2025), we reconstruct 102.2M career works by 6.15M contributors, and for the 908k contributors whose repositories publish packages, we measure each work's impact by how many downstream packages come to depend on it. First, we find that the timing of a career's biggest hit is close to a lottery over their works, as in science and the arts, with a small, replicable lean toward early career that grows as careers get longer. Second, some coders reliably produce higher-impact work than others, but this lasting personal factor accounts for only part of why impact persists (about a fifth in our primary specification); the rest behaves like momentum, success feeding on itself for a period of time. Third, within the same contributors, this structure did not change after ChatGPT's release. The stable factor's weight grew by about as much as it grew for an earlier cohort that simply aged, and subtracting the effect of aging from the effect of generative AI puts the shift at +0.03 (95% CI [-0.22, +0.23]), indistinguishable from zero. The success pattern documented in science and the arts therefore describes open-source careers too, and it shows no detectable break across the arrival of generative AI. These results have implications for how track records on open platforms should be read and on what to expect from generative AI for the careers built on them.

cs.SI

How neighbourhood ideology shapes misinformation belief in densely tied social networks

With the rapid spread of news on social media, understanding the propagation of misinformation is becoming increasingly important. One factor that affects individuals' vulnerability to false information is their ideological predisposition. Despite the large number of agent-based models that focus on social influence as a driver of the spread of false claims, they often fail to explicitly integrate personal ideological biases into belief formation. In this work, we explore how misinformation spreads through the interaction between individuals' ideological biases and social influence. Our model accounts for both the strength of individuals' ideological biases and the extent to which a false claim aligns with their ideology. Social influence modifies the effects of ideological intensity and false claim alignment through network interactions. Notably, the influence of neighbours' ideological intensity on belief is strongly affected by how well those neighbours are connected to one another. These results highlight the importance of considering both network structure and personal ideological biases when modelling misinformation propagation.

cs.SI