Tootfinder

Mitigating Strategy Preference Bias in Emotional Support Conversation via Uncertainty Estimations
Yougen Zhou, Qin Chen, Ningning Zhou, Jie Zhou, Xingjiao Wu, Liang He
https://arxiv.org/abs/2509.12661 …

Mitigating Strategy Preference Bias in Emotional Support Conversation via Uncertainty Estimations
Emotional support conversation (ESC) aims to alleviate distress through empathetic dialogue, yet large language models (LLMs) face persistent challenges in delivering effective ESC due to low accuracy in strategy planning. Moreover, there is a considerable preference bias towards specific strategies. Prior methods using fine-tuned strategy planners have shown potential in reducing such bias, while the underlying causes of the preference bias in LLMs have not well been studied. To address these …

Do Large Language Models Favor Recent Content? A Study on Recency Bias in LLM-Based Reranking
Hanpei Fang, Sijie Tao, Nuo Chen, Kai-Xin Chang, Tetsuya Sakai
https://arxiv.org/abs/2509.11353

Do Large Language Models Favor Recent Content? A Study on Recency Bias in LLM-Based Reranking
Large language models (LLMs) are increasingly deployed in information systems, including being used as second-stage rerankers in information retrieval pipelines, yet their susceptibility to recency bias has received little attention. We investigate whether LLMs implicitly favour newer documents by prepending artificial publication dates to passages in the TREC Deep Learning passage retrieval collections in 2021 (DL21) and 2022 (DL22). Across seven models, GPT-3.5-turbo, GPT-4o, GPT-4, LLaMA-3 8…

Least squares-based methods to bias adjustment in scalar-on-function regression model using a functional instrumental variable
Xiwei Chen, Ufuk Beyaztas, Caihong Qin, Heyang Ji, Gilson Honvoh, Roger S. Zoh, Lan Xue, Carmen D. Tekwe
https://arxiv.org/abs/2509.12122

Least squares-based methods to bias adjustment in scalar-on-function regression model using a functional instrumental variable
Instrumental variables are widely used to adjust for measurement error bias when assessing associations of health outcomes with ME prone independent variables. IV approaches addressing ME in longitudinal models are well established, but few methods exist for functional regression. We develop two methods to adjust for ME bias in scalar on function linear models. We regress a scalar outcome on an ME prone functional variable using a functional IV for model identification and propose two least squ…

Noch einige der zuletzt hier besonders häufig geteilten #News:
Trumps anti-woke-Verordnung: Meta arbeitet mit rechtem Influencer

Bias in the tensor-to-scalar ratio from self-interacting dark radiation
Nahuel Mir\'on-Granese, Claudia G. Sc\'occola
https://arxiv.org/abs/2509.10607 https://

Bias in the tensor-to-scalar ratio from self-interacting dark radiation
We investigate the cosmological imprint of self-interacting dark radiation (DR) on the primordial $B$-mode angular power spectrum and its impact on the estimation of the tensor-to-scalar ratio $r$. We consider a minimal model in which DR is described as an effectively massless axion-like particle with quartic self-interactions. These interactions are incorporated into the Einstein-Boltzmann equations using the relaxation time approximation and implemented in the $\texttt{CLASS}$ code. We show t…

Universal relations for fast rotating neutron stars without equation of state bias
Christian J. Kr\"uger, Mariachiara Celato
https://arxiv.org/abs/2509.11882 https://

Universal relations for fast rotating neutron stars without equation of state bias
We provide a summary of several parametrisations for the nuclear equation of state that have been proposed over the past decades and list viable ranges for their parameters. Based on these parametrisations, we construct a large database of rotating neutron star models, which are arranged in sequences of constant central energy density. After filtering these sequences with respect to generous astrophysical constraints, we discover tight universal relations for various bulk quantities of uniforml…

JustEva: A Toolkit to Evaluate LLM Fairness in Legal Knowledge Inference
Zongyue Xue, Siyuan Zheng, Shaochun Wang, Yiran Hu, Shenran Wang, Yuxin Yao, Haitao Li, Qingyao Ai, Yiqun Liu, Yun Liu, Weixing Shen
https://arxiv.org/abs/2509.12104

JustEva: A Toolkit to Evaluate LLM Fairness in Legal Knowledge Inference
The integration of Large Language Models (LLMs) into legal practice raises pressing concerns about judicial fairness, particularly due to the nature of their "black-box" processes. This study introduces JustEva, a comprehensive, open-source evaluation toolkit designed to measure LLM fairness in legal tasks. JustEva features several advantages: (1) a structured label system covering 65 extra-legal factors; (2) three core fairness metrics - inconsistency, bias, and imbalanced inaccuracy; (3) robu…

AesBiasBench: Evaluating Bias and Alignment in Multimodal Language Models for Personalized Image Aesthetic Assessment
Kun Li, Lai-Man Po, Hongzheng Yang, Xuyuan Xu, Kangcheng Liu, Yuzhi Zhao
https://arxiv.org/abs/2509.11620

AesBiasBench: Evaluating Bias and Alignment in Multimodal Language Models for Personalized Image Aesthetic Assessment
Multimodal Large Language Models (MLLMs) are increasingly applied in Personalized Image Aesthetic Assessment (PIAA) as a scalable alternative to expert evaluations. However, their predictions may reflect subtle biases influenced by demographic factors such as gender, age, and education. In this work, we propose AesBiasBench, a benchmark designed to evaluate MLLMs along two complementary dimensions: (1) stereotype bias, quantified by measuring variations in aesthetic evaluations across demograph…

Machine Unlearning for Responsible and Adaptive AI in Education
Betty Mayeku, Sandra Hummel, Parisa Memarmoshrefi
https://arxiv.org/abs/2509.10590 https://…

Machine Unlearning for Responsible and Adaptive AI in Education
The concept of Machine Unlearning (MU) has gained popularity in various domains due to its ability to address several issues in Machine Learning (ML) models, particularly those related to privacy, security, bias mitigation, and adaptability. With these abilities, MU is evolving into a promising technology in upholding Responsible AI principles and optimizing ML models' performance. However, despite its promising potential, the concept has not received much attention in the education sector. In …

Task-Agnostic Learnable Weighted-Knowledge Base Scheme for Robust Semantic Communications
Shiyao Jiang, Jian Jiao, Xingjian Zhang, Ye Wang, Dusit Niyato, Qinyu Zhang
https://arxiv.org/abs/2509.11636

Task-Agnostic Learnable Weighted-Knowledge Base Scheme for Robust Semantic Communications
With the emergence of diverse and massive data in the upcoming sixth-generation (6G) networks, the task-agnostic semantic communication system is regarded to provide robust intelligent services. In this paper, we propose a task-agnostic learnable weighted-knowledge base semantic communication (TALSC) framework for robust image transmission to address the real-world heterogeneous data bias in KB, including label flipping noise and class imbalance. The TALSC framework incorporates a sample confid…

Es wurde ja erwartet, aber tatsächlich fand ich den Punkt der Verstärkung bestehender Ungleichheiten bislang stark unterrepräsentiert in der Berichterstattung.
KI-Bias bei Gehaltsratschlägen: Wie Sprachmodelle Frauen systematisch benachteiligen - Notebookcheck.com News

Crosslisted article(s) found for hep-ph. https://arxiv.org/list/hep-ph/new
[1/1]:
- Bias in the tensor-to-scalar ratio from self-interacting dark radiation
Nahuel Mir\'on-Granese, Claudia G. Sc\'occola

Prediction-Powered Inference with Inverse Probability Weighting
Jyotishka Datta, Nicholas G. Polson
https://arxiv.org/abs/2508.10149 https://arxiv.org/pdf/…

Prediction-Powered Inference with Inverse Probability Weighting
Prediction-powered inference (PPI) is a recent framework for valid statistical inference with partially labeled data, combining model-based predictions on a large unlabeled set with bias correction from a smaller labeled subset. We show that PPI can be extended to handle informative labeling by replacing its unweighted bias-correction term with an inverse probability weighted (IPW) version, using the classical Horvitz--Thompson or Hájek forms. This connection unites design-based survey samplin…

Replaced article(s) found for econ.EM. https://arxiv.org/list/econ.EM/new
[1/1]:
- Teacher bias or measurement error?
Thomas van Huizen, Madelon Jacobs, Matthijs Oosterveen

Racial bias, colorism, and overcorrection
Kenneth Colombe, Alex Krumer, Rosa Lavelle-Hill, Tim Pawlowski
https://arxiv.org/abs/2508.10585 https://arxiv.org…

Racial bias, colorism, and overcorrection
This paper examines whether increased awareness can affect racial bias and colorism. We exploit a natural experiment from the widespread publicity of Price and Wolfers (2010), which intensified scrutiny of racial bias in men's basketball officiating. We investigate refereeing decisions in the Women's National Basketball Association (WNBA), an organization with a long-standing commitment to diversity, equity, and inclusion (DEI). We apply machine learning techniques to predict player race and to…

In response to Elon Musk's claims that the App Store favors the ChatGPT app, Apple says the App Store "is designed to be fair and free of bias" (Mark Gurman/@markgurman)
https://x.com/markgurman/status/1955383759853007198

Mark Gurman (@markgurman) on X
Apple on @elonmusk’s claims: "The App Store is designed to be fair and free of bias. We feature thousands of apps through charts, algorithmic recommendations, and curated lists selected by experts using objective criteria.” https://t.co/w4NFINOF9x

Trumps anti-woke-Verordnung: Meta arbeitet mit rechtem Influencer
Um den vermeintlich woken Bias aus KI-Anwendungen zu bekommen, arbeitet Meta nun mit dem konservativen Robby Starbuck zusammen.
https://www.…

Scalable Training for Vector-Quantized Networks with 100% Codebook Utilization
Yifan Chang, Jie Qin, Limeng Qiao, Xiaofeng Wang, Zheng Zhu, Lin Ma, Xingang Wang
https://arxiv.org/abs/2509.10140

Scalable Training for Vector-Quantized Networks with 100% Codebook Utilization
Vector quantization (VQ) is a key component in discrete tokenizers for image generation, but its training is often unstable due to straight-through estimation bias, one-step-behind updates, and sparse codebook gradients, which lead to suboptimal reconstruction performance and low codebook usage. In this work, we analyze these fundamental challenges and provide a simple yet effective solution. To maintain high codebook usage in VQ networks (VQN) during learning annealing and codebook size expans…

Distinguishing Majorana bound states from accidental zero-energy modes with a microwave cavity
Sarath Prem, Olesia Dmytruk, Mircea Trif
https://arxiv.org/abs/2509.13194 https://…

Distinguishing Majorana bound states from accidental zero-energy modes with a microwave cavity
Transport measurements of hybrid nanowires often rely on the observation of a zero-bias conductance peak as a hallmark of Majorana bound states (MBSs). However, such signatures can also be produced by trivial zero-energy Andreev bound states (ABSs) or by quasi-Majorana bound states (QMBSs), complicating their unambiguous identification. Here we propose microwave absorption visibility, extracted from parity-dependent cavity-nanowire susceptibility measurements, as a complementary probe of MBSs n…

Stabilizing Long-term Multi-turn Reinforcement Learning with Gated Rewards
Zetian Sun, Dongfang Li, Zhuoen Chen, Yuhuai Qin, Baotian Hu
https://arxiv.org/abs/2508.10548 https://…

Stabilizing Long-term Multi-turn Reinforcement Learning with Gated Rewards
Reward sparsity in long-horizon reinforcement learning (RL) tasks remains a significant challenge, while existing outcome-based reward shaping struggles to define meaningful immediate rewards without introducing bias or requiring explicit task decomposition. Alternatively, verification-based reward shaping uses stepwise critics, but misalignment between immediate rewards and long-term objectives can lead to reward hacking and suboptimal policies. In this work, we address this problem in the con…

Don't Change My View: Ideological Bias Auditing in Large Language Models
Paul Kr\"oger, Emilio Barkett
https://arxiv.org/abs/2509.12652 https://ar…

Don't Change My View: Ideological Bias Auditing in Large Language Models
As large language models (LLMs) become increasingly embedded in products used by millions, their outputs may influence individual beliefs and, cumulatively, shape public opinion. If the behavior of LLMs can be intentionally steered toward specific ideological positions, such as political or religious views, then those who control these systems could gain disproportionate influence over public discourse. Although it remains an open question whether LLMs can reliably be guided toward coherent ide…

LLM-Based Instance-Driven Heuristic Bias In the Context of a Biased Random Key Genetic Algorithm
Camilo Chac\'on Sartori, Mart\'in Isla Pino, Pedro Pinacho-Davidson, Christian Blum
https://arxiv.org/abs/2509.09707

LLM-Based Instance-Driven Heuristic Bias In the Context of a Biased Random Key Genetic Algorithm
Integrating Large Language Models (LLMs) within metaheuristics opens a novel path for solving complex combinatorial optimization problems. While most existing approaches leverage LLMs for code generation to create or refine specific heuristics, they often overlook the structural properties of individual problem instances. In this work, we introduce a novel framework that integrates LLMs with a Biased Random-Key Genetic Algorithm (BRKGA) to solve the NP-hard Longest Run Subsequence problem. Our …

The Social Cost of Intelligence: Emergence, Propagation, and Amplification of Stereotypical Bias in Multi-Agent Systems
Thi-Nhung Nguyen, Linhao Luo, Thuy-Trang Vu, Dinh Phung
https://arxiv.org/abs/2510.10943

The Social Cost of Intelligence: Emergence, Propagation, and Amplification of Stereotypical Bias in Multi-Agent Systems
Bias in large language models (LLMs) remains a persistent challenge, manifesting in stereotyping and unfair treatment across social groups. While prior research has primarily focused on individual models, the rise of multi-agent systems (MAS), where multiple LLMs collaborate and communicate, introduces new and largely unexplored dynamics in bias emergence and propagation. In this work, we present a comprehensive study of stereotypical bias in MAS, examining how internal specialization, underlyi…

Towards a precise measurement of the $\Lambda_c^ /D^0$ ratio at RHIC
Joseph D. Osborn
https://arxiv.org/abs/2509.10772 https://arxiv.org/pdf/2509.10772

Towards a precise measurement of the $Λ_c^+/D^0$ ratio at RHIC
sPHENIX is a next-generation experiment at RHIC for jet and heavy-flavor physics which was fully commissioned during 2023 and 2024. Using its novel streaming-readout-capable, precision tracking system, sPHENIX collected 100 billion unbiased $p$$+$$p$ collisions, and a further sample of minimum-bias Au-Au collisions, in Run-24. A key measurement of the sPHENIX heavy flavor physics program is the comparison of $Λ_c^+$ to $D^0$ differential yields in both Au+Au and $p$$+$$p$ collisions, which pro…

Systematic Schrieffer-Wolff-transformation approach to Josephson junctions: quasiparticle effects and Josephson harmonics
\'Ad\'am B\'acsi, Teodor Ili\v{c}in, Rok \v{Z}itko
https://arxiv.org/abs/2509.12706

Systematic Schrieffer-Wolff-transformation approach to Josephson junctions: quasiparticle effects and Josephson harmonics
We use the Schrieffer-Wolff transformation (SWT) to analyze Josephson junctions between superconducting leads described by the charge-conserving BCS theory. Starting from the single-electron tunneling terms, we directly recover the conventional effective Hamiltonian, $-E_J\cos\hatφ$, with an operator-valued phase bias $\hatφ$. The SWT approach has the advantage that it can be systematically extended to more complex scenarios. We show that if a Bogoliubov quasiparticle is present its motion co…

Ein Artikel in Plos One beleuchtet den politischen #Bias von #Chatbots und zeigt, dass viele KI-Modelle eher linksgerichtete Antworten geben.
Dies könnte auf das Supervised Fine-Tuning zurückzuführen sein, bei dem KI durch menschliche Beispiele lernt. Interessanterweise weisen unterschiedliche

Linksgerichtete Chatbots? Eine Analyse der politischen Ausrichtung von LLMs
Politischer Bias in ChatGPT und Co.: Gibt es eine politische Ausrichtung in LLMs und wodurch könnte sie entstehen?

😕 Why late surges in election results fuel fraud suspicions among voters
#elections

Why late surges in election results fuel fraud suspicions among voters
Once an electoral candidate is in the lead, the public views them as the winner. If the candidate then loses, fraud seems a likely explanation, and even the winner's supporters might be inclined to be suspicious about the final result. This is because of the cumulative redundancy bias (CRB).

Based AI improves human decision-making but reduces trust
Shiyang Lai, Junsol Kim, Nadav Kunievsky, Yujin Potter, James Evans
https://arxiv.org/abs/2508.09297 https://

Based AI improves human decision-making but reduces trust
Current AI systems minimize risk by enforcing ideological neutrality, yet this may introduce automation bias by suppressing cognitive engagement in human decision-making. We conducted randomized trials with 2,500 participants to test whether culturally biased AI enhances human decision-making. Participants interacted with politically diverse GPT-4o variants on information evaluation tasks. Partisan AI assistants enhanced human performance, increased engagement, and reduced evaluative bias compa…

Evaluating and Mitigating LLM-as-a-judge Bias in Communication Systems
Jiaxin Gao, Chen Chen, Yanwen Jia, Xueluan Gong, Kwok-Yan Lam, Qian Wang
https://arxiv.org/abs/2510.12462 …

Evaluating and Mitigating LLM-as-a-judge Bias in Communication Systems
Large Language Models (LLMs) are increasingly being used to autonomously evaluate the quality of content in communication systems, e.g., to assess responses in telecom customer support chatbots. However, the impartiality of these AI "judges" is not guaranteed, and any biases in their evaluation criteria could skew outcomes and undermine user trust. In this paper, we systematically investigate judgment biases in two LLM-as-a-judge models (i.e., GPT-Judge and JudgeLM) under the point-wise scoring…

A Latent Factor Panel Approach to Spatiotemporal Causal Inference
Jiaxi Wu, Alexander Franks
https://arxiv.org/abs/2509.10974 https://arxiv.org/pdf/2509.10…

A Latent Factor Panel Approach to Spatiotemporal Causal Inference
Unmeasured confounding can severely bias causal effect estimates from spatiotemporal observational data, especially when the confounders do not vary smoothly in time and space. In this work, we develop a method for addressing unmeasured confounding in spatiotemporal contexts by building on models from the panel data literature and methods in multivariate causal inference. Our method is based on a factor confounding assumption, which posits that effects of unmeasured confounders on exposures and…

Crosslisted article(s) found for cs.SD. https://arxiv.org/list/cs.SD/new
[1/1]:
- Not in Sync: Unveiling Temporal Bias in Audio Chat Models
Jiayu Yao, Shenghua Liu, Yiwei Wang, Rundong Cheng, Lingrui Mei, Baolong Bi, Zhen Xiong, Xueqi Cheng

Inductive Bias Extraction and Matching for LLM Prompts
Christian M. Angel, Francis Ferraro
https://arxiv.org/abs/2508.10295 https://arxiv.org/pdf/2508.1029…

Inductive Bias Extraction and Matching for LLM Prompts
The active research topic of prompt engineering makes it evident that LLMs are sensitive to small changes in prompt wording. A portion of this can be ascribed to the inductive bias that is present in the LLM. By using an LLM's output as a portion of its prompt, we can more easily create satisfactory wording for prompts. This has the effect of creating a prompt that matches the inductive bias in model. Empirically, we show that using this Inductive Bias Extraction and Matching strategy improves …

Correcting for partial verification bias in diagnostic accuracy studies: A tutorial using R
Wan Nor Arifin, Umi Kalsom Yusof
https://arxiv.org/abs/2509.12217 https://

Correcting for partial verification bias in diagnostic accuracy studies: A tutorial using R
Diagnostic tests play a crucial role in medical care. Thus any new diagnostic tests must undergo a thorough evaluation. New diagnostic tests are evaluated in comparison with the respective gold standard tests. The performance of binary diagnostic tests is quantified by accuracy measures, with sensitivity and specificity being the most important measures. In any diagnostic accuracy study, the estimates of these measures are often biased owing to selective verification of the patients, which is r…

Exploring Content and Social Connections of Fake News with Explainable Text and Graph Learning
V\'itor N. Louren\c{c}o, Aline Paes, and Tillman Weyde
https://arxiv.org/abs/2508.10040

Exploring Content and Social Connections of Fake News with Explainable Text and Graph Learning
The global spread of misinformation and concerns about content trustworthiness have driven the development of automated fact-checking systems. Since false information often exploits social media dynamics such as "likes" and user networks to amplify its reach, effective solutions must go beyond content analysis to incorporate these factors. Moreover, simply labelling content as false can be ineffective or even reinforce biases such as automation and confirmation bias. This paper proposes an expl…

From Delegates to Trustees: How Optimizing for Long-Term Interests Shapes Bias and Alignment in LLM
Suyash Fulay, Jocelyn Zhu, Michiel Bakker
https://arxiv.org/abs/2510.12689 ht…

From Delegates to Trustees: How Optimizing for Long-Term Interests Shapes Bias and Alignment in LLM
Large language models (LLMs) have shown promising accuracy in predicting survey responses and policy preferences, which has increased interest in their potential to represent human interests in various domains. Most existing research has focused on behavioral cloning, effectively evaluating how well models reproduce individuals' expressed preferences. Drawing on theories of political representation, we highlight an underexplored design trade-off: whether AI systems should act as delegates, mirr…

✅ double-checking PhD review for a female candidate that I didn't fall into the gender bias trap #academicChatter #amReviewing #effortVsAccomplishment #skillVsGrindstone

Realizing Parrondo's Paradox in Single-Qubit Quantum Walks via Local Phase-Induced Spatial Inhomogeneity
Ran-Yu Chang, Yun-Hsuan Chen, Gooi Zi Liang, Tsung-Wei Huang
https://arxiv.org/abs/2508.09457

Realizing Parrondo's Paradox in Single-Qubit Quantum Walks via Local Phase-Induced Spatial Inhomogeneity
Parrondo's paradox describes a counterintuitive phenomenon where alternating between two individually losing games results in a winning expectation. While its classical origin relies on capital-dependent bias and noise-induced asymmetry, realizing a robust quantum version of the paradox has remained challenging, especially under the constraint of single-qubit coin systems. In this work, we demonstrate that a genuine quantum Parrondo effect can emerge in discrete-time quantum walks (DTQWs) by al…

Delayed phase mixing in the self-gravitating Galactic disc
T. Asano, T. Antoja
https://arxiv.org/abs/2510.11801 https://arxiv.org/pdf/2510.11801

Delayed phase mixing in the self-gravitating Galactic disc
The Gaia phase spiral is considered to work as a dynamical clock for dating past perturbations, but most of the previous studies neglected the disc's self-gravity, potentially biasing estimates of the phase spiral's excitation time. We aim to evaluate the impact of self-gravitating effects on the evolution of vertical phase spirals and to quantify the bias introduced in estimating their excitation time when such effects are ignored. We analysed a high-resolution, self-consistent $N$-body simula…

…oooohhhhh this s____t FINALLY might start hitting the fan! I, and many others, have been pointing out the monopoly that BC company ConAir has over firefighting contracts for Canadian governments for years.
Coulson Aviation has been doing firefighting aircraft for decades and has never been picked by its own home province on a long term contract.
The bias has been glaring.
Looks like they're going to fight it out in Court in Saskatchewan!
#BCWildfire #BCPoli
https://www.cbc.ca/news/canada/saskatchewan/government-overpaid-firefighting-planes-manufacturer-says-1.7630608

General model and modulation strategies for Sagnac-based encoders
Federico Berra, Mat\'ias Rub\'en Bola\~nos, Alberto De Toni, Kannan Vijayadharan, Costantino Agnesi, Marco Avesani, Andrea Stanco, Paolo Villoresi, Giuseppe Vallone
https://arxiv.org/abs/2510.11873

General model and modulation strategies for Sagnac-based encoders
In recent decades, there has been an increasing demand for faster modulation schemes. Electro-optic modulators are essential components in modern photonic systems, enabling high-speed control of light for applications ranging from telecommunications to quantum communication. Conventional inline and Mach-Zehnder modulators, while widely adopted, are limited by bias drift, high operating voltages, and polarization-mode dispersion. Sagnac loop-based modulators have recently emerged as a promising …

How the US democracy is designed to avoid representation
Right now in the US, a system which proclaims to give each citizen representation, my interests are not represented very well by most of my so-called representatives at any level of government. This is true for a majority of Americans across the political spectrum, and it happens by design. The "founding fathers" were explicit about wanting a system of government that would appear Democratic but which would keep power in the hands of rich white landowners, and they successfully designed exactly that. But how does disenfranchisement work in this system?
First, a two-party system locked in by first-post-the-post winner-takes-all elections immediately destroys representation for everyone who didn't vote for the winner, including those who didn't vote or weren't eligible to vote. Single-day non-holiday elections and prisoner disenfranchisement go a long way towards ensuring working-class people get no say, but much larger is the winner-takes all system. In fact, even people who vote for the winning candidate don't get effective representation if they're really just voting against the opponent as the greater of two evils. In a 51/49 election with 50% turnout, you've immediately ensured that ~75% of eligible voters don't get represented, and with lesser-of-two-evils voting, you create an even wider gap to wedge corporate interests into. Politicians need money to saturate their lesser-of-two-evils message far more than they need to convince any individual voter to support their policies. It's even okay if they get caught lying, cheating, or worse (cough Epstein cough) as long as the other side is also doing those things and you can freeze out new parties.
Second, by design the Senate ensures uneven representation, allowing control of the least-populous half of states to control or at least shut down the legislative process. A rough count suggests 284.6 million live in the 25 most-populous states, while only 54.8 million live in the rest. Currently, counting states with divided representation as two half-states with half as much population, 157.8 million people are represented by 53 Republican sensors, while 180.5 million people get only 45 seats of Democratic representation. This isn't an anti-Democrat bias, it's a bias towards less-populous states, whose residents get more than their share it political power.
I haven't even talked about gerrymandering yet, or family/faith-based "party loyalty," etc. Overall, the effect is that the number of people whose elected representatives meaningfully represent their interests on any given issue is vanishingly small (like, 10% of people tops), unless you happen to be rich enough to purchase lobbying power or direct access.
If we look at polls, we can see how lack of representation lets congress & the president enact many policies that go against what a majority of the population wants. Things like abortion restrictions, the current ICE raids, and Medicare cuts are deeply unpopular, but they benefit the political class and those who can buy access. These are possible because the system ensures at every step of the way that ordinary people do NOT get the one thing the system promises them: representation in the halls of power.
Okay, but is this a feature of all democracies, inherent in the nature of a majority-decides system? Not exactly...
1/2
#uspol #democracy

Mask Consistency Regularization in Object Removal
Hua Yuan, Jin Yuan, Yicheng Jiang, Yao Zhang, Xin Geng, Yong Rui
https://arxiv.org/abs/2509.10259 https://

Mask Consistency Regularization in Object Removal
Object removal, a challenging task within image inpainting, involves seamlessly filling the removed region with content that matches the surrounding context. Despite advancements in diffusion models, current methods still face two critical challenges. The first is mask hallucination, where the model generates irrelevant or spurious content inside the masked region, and the second is mask-shape bias, where the model fills the masked area with an object that mimics the mask's shape rather than su…

On Spectral Properties of Gradient-based Explanation Methods
Amir Mehrpanah, Erik Englesson, Hossein Azizpour
https://arxiv.org/abs/2508.10595 https://arxi…

On Spectral Properties of Gradient-based Explanation Methods
Understanding the behavior of deep networks is crucial to increase our confidence in their results. Despite an extensive body of work for explaining their predictions, researchers have faced reliability issues, which can be attributed to insufficient formalism. In our research, we adopt novel probabilistic and spectral perspectives to formally analyze explanation methods. Our study reveals a pervasive spectral bias stemming from the use of gradient, and sheds light on some common design choices…

Mutual synchronization of two asymmetric-nano-constriction-based spin-Hall nano-oscillators
Roman V. Ovcharov, Roman S. Khymyn, Akash Kumar, Johan \r{A}kerman
https://arxiv.org/abs/2509.12113

Mutual synchronization of two asymmetric-nano-constriction-based spin-Hall nano-oscillators
We propose an asymmetric-nanoconstriction (ANC) design of spin-Hall nano-oscillators (SHNOs) and investigate mutual synchronization of a pair of such devices using micromagnetic simulations. The ANC geometry enables strong dipolar coupling at sub-50 nm separations while preserving independent current bias for each oscillator. We first characterize the auto-oscillation of a single ANC-SHNO, revealing a broad frequency tuning range and a field-controlled crossover between negative and positive no…

FuXi-\beta: Towards a Lightweight and Fast Large-Scale Generative Recommendation Model
Yufei Ye, Wei Guo, Hao Wang, Hong Zhu, Yuyang Ye, Yong Liu, Huifeng Guo, Ruiming Tang, Defu Lian, Enhong Chen
https://arxiv.org/abs/2508.10615

FuXi-β: Towards a Lightweight and Fast Large-Scale Generative Recommendation Model
Scaling laws for autoregressive generative recommenders reveal potential for larger, more versatile systems but mean greater latency and training costs. To accelerate training and inference, we investigated the recent generative recommendation models HSTU and FuXi-$α$, identifying two efficiency bottlenecks: the indexing operations in relative temporal attention bias and the computation of the query-key attention map. Additionally, we observed that relative attention bias in self-attention mec…

Using Oral Exams in Physics and Astronomy Courses
Brian DiGiorgio Zanger
https://arxiv.org/abs/2509.09846 https://arxiv.org/pdf/2509.09846

Using Oral Exams in Physics and Astronomy Courses
Oral exams were common historically across academia, though their popularity has recently fallen. Many argue against them as an assessment technique because they are vulnerable to bias and subjectivity, difficult to administer, and impractical in large college classes. I present a method for administering oral examinations in upper-level courses that mitigates some of the disadvantages. This method creates a rigid question structure meant to assess student mastery of material, have a well-defin…

Power-Dominance in Estimation Theory: A Third Pathological Axis
Sri Satish Krishna Chaitanya Bulusu, Mikko Sillanp\"a\"a
https://arxiv.org/abs/2509.12691 https://

Power-Dominance in Estimation Theory: A Third Pathological Axis
This paper introduces a novel framework for estimation theory by introducing a second-order diagnostic for estimator design. While classical analysis focuses on the bias-variance trade-off, we present a more foundational constraint. This result is model-agnostic, domain-agnostic, and is valid for both parametric and non-parametric problems, Bayesian and frequentist frameworks. We propose to classify the estimators into three primary power regimes. We theoretically establish that any estimator o…

Analysing Moral Bias in Finetuned LLMs through Mechanistic Interpretability
Bianca Raimondi, Daniela Dalbagno, Maurizio Gabbrielli
https://arxiv.org/abs/2510.12229 https://

Analysing Moral Bias in Finetuned LLMs through Mechanistic Interpretability
Large language models (LLMs) have been shown to internalize human-like biases during finetuning, yet the mechanisms by which these biases manifest remain unclear. In this work, we investigated whether the well-known Knobe effect, a moral bias in intentionality judgements, emerges in finetuned LLMs and whether it can be traced back to specific components of the model. We conducted a Layer-Patching analysis across 3 open-weights LLMs and demonstrated that the bias is not only learned during finet…

An investigation finds rampant caste bias in ChatGPT and Sora; a researcher also finds caste bias in Sarvam AI, which touts itself as a sovereign AI for India (Nilesh Christopher/MIT Technology Review)
https://www.technologyreview.com/2025/10/01/1124621/openai-in…

OpenAI is huge in India. Its models are steeped in caste bias.
India is OpenAI’s second-largest market, but ChatGPT and Sora reproduce caste stereotypes that harm millions of people.

Closed-form parameter estimation for the bivariate gamma distribution: New approaches
Roberto Vila, Helton Saulo
https://arxiv.org/abs/2509.10794 https://a…

Closed-form parameter estimation for the bivariate gamma distribution: New approaches
We propose new closed-form estimators for the parameters of McKay's bivariate gamma distribution by exploiting monotone transformations of the likelihood equations. As a special case, our framework recovers the estimators recently introduced by Zhao et al. (2022) [Zhao, J., Jang, Y.-H., and Kim, H. (2022). Closed-form and bias-corrected estimators for the bivariate gamma distribution. Journal of Multivariate Analysis, 191:105009]. Theoretical properties, including strong consistency and asympto…

Mitigating the Noise Shift for Denoising Generative Models via Noise Awareness Guidance
Jincheng Zhong, Boyuan Jiang, Xin Tao, Pengfei Wan, Kun Gai, Mingsheng Long
https://arxiv.org/abs/2510.12497

Mitigating the Noise Shift for Denoising Generative Models via Noise Awareness Guidance
Existing denoising generative models rely on solving discretized reverse-time SDEs or ODEs. In this paper, we identify a long-overlooked yet pervasive issue in this family of models: a misalignment between the pre-defined noise level and the actual noise level encoded in intermediate states during sampling. We refer to this misalignment as noise shift. Through empirical analysis, we demonstrate that noise shift is widespread in modern diffusion models and exhibits a systematic bias, leading to …

Positional Encoding via Token-Aware Phase Attention
Yu (Sid), Wang, Sheng Shen, R\'emi Munos, Hongyuan Zhan, Yuandong Tian
https://arxiv.org/abs/2509.12635 https://

Positional Encoding via Token-Aware Phase Attention
We prove under practical assumptions that Rotary Positional Embedding (RoPE) introduces an intrinsic distance-dependent bias in attention scores that limits RoPE's ability to model long-context. RoPE extension methods may alleviate this issue, but they typically require post-hoc adjustments after pretraining, such as rescaling or hyperparameters retuning. This paper introduces Token-Aware Phase Attention (TAPA), a new positional encoding method that incorporates a learnable phase function into …

A hierarchical entropy method for the delocalization of bias in high-dimensional Langevin Monte Carlo
Daniel Lacker, Fuzhong Zhou
https://arxiv.org/abs/2509.08619 https://

A hierarchical entropy method for the delocalization of bias in high-dimensional Langevin Monte Carlo
The unadjusted Langevin algorithm is widely used for sampling from complex high-dimensional distributions. It is well known to be biased, with the bias typically scaling linearly with the dimension when measured in squared Wasserstein distance. However, the recent paper of Chen et al. (2024) identifies an intriguing new delocalization effect: For a class of distributions with sparse interactions, the bias between low-dimensional marginals scales only with the lower dimension, not the full dimen…

UDA: Unsupervised Debiasing Alignment for Pair-wise LLM-as-a-Judge
Yang Zhang, Cunxiang Wang, Lindong Wu, Wenbo Yu, Yidong Wang, Guangsheng Bao, Jie Tang
https://arxiv.org/abs/2508.09724

UDA: Unsupervised Debiasing Alignment for Pair-wise LLM-as-a-Judge
Pairwise evaluation of Large Language Models (LLMs) is a common paradigm, but it is prone to preference bias, where judges systematically favor certain outputs, such as their own. This bias leads to inconsistent and skewed rankings across different judges. To address this, we first empirically demonstrate significant and heterogeneous biases in cross-model evaluations. We then propose UDA (Unsupervised Debiasing Alignment), a framework that reduces inter-judge disagreement by dynamically adjust…

Hidden Figures of Globular Clusters: Integrated Stellar Populations Impacted by Hot Subdwarfs
Thayse A. Pacheco, Paula R. T. Coelho, Lucimara P. Martins, Ricardo P. Schiavon, Erik V. R. de Lima, Marcos P. Diaz, Domenico Nardiello, Ronaldo S. Levenhagen, Rogerio Riffel, Charles J. Bonatto, Ana L. Chies-Santos
https://arxiv.org/abs/2510.1232…

Hidden Figures of Globular Clusters: Integrated Stellar Populations Impacted by Hot Subdwarfs
Globular clusters (GCs) are fundamental for understanding the integrated light of old stellar populations and galaxy assembly processes. However, the role of hot, evolved stars, such as horizontal branch (HB), extreme HB, and blue stragglers, remains poorly constrained. These stars are often underrepresented or entirely excluded from stellar population models, despite their dominant contribution to the ultraviolet (UV) flux. Their presence can bias age estimates by mimicking the spectral signat…

Bias-Aware AI Chatbot for Engineering Advising at the University of Maryland A. James Clark School of Engineering
Prarthana P. Kartholy, Thandi M. Labor, Neil N. Panchal, Sean H. Wang, Hillary N. Owusu
https://arxiv.org/abs/2510.09636

Bias-Aware AI Chatbot for Engineering Advising at the University of Maryland A. James Clark School of Engineering
Selecting a college major is a difficult decision for many incoming freshmen. Traditional academic advising is often hindered by long wait times, intimidating environments, and limited personalization. AI Chatbots present an opportunity to address these challenges. However, AI systems also have the potential to generate biased responses, prejudices related to race, gender, socioeconomic status, and disability. These biases risk turning away potential students and undermining reliability of AI s…

Network Traffic as a Scalable Ethnographic Lens for Understanding University Students' AI Tool Practices
Donghan Hu, Rameen Mahmood, Annabelle David, Danny Yuxing Huang
https://arxiv.org/abs/2510.09763

Network Traffic as a Scalable Ethnographic Lens for Understanding University Students' AI Tool Practices
AI-driven applications have become woven into students' academic and creative workflows, influencing how they learn, write, and produce ideas. Gaining a nuanced understanding of these usage patterns is essential, yet conventional survey and interview methods remain limited by recall bias, self-presentation effects, and the underreporting of habitual behaviors. While ethnographic methods offer richer contextual insights, they often face challenges of scale and reproducibility. To bridge this gap…

EthicsMH: A Pilot Benchmark for Ethical Reasoning in Mental Health AI
Sai Kartheek Reddy Kasu
https://arxiv.org/abs/2509.11648 https://arxiv.org/pdf/2509.1…

EthicsMH: A Pilot Benchmark for Ethical Reasoning in Mental Health AI
The deployment of large language models (LLMs) in mental health and other sensitive domains raises urgent questions about ethical reasoning, fairness, and responsible alignment. Yet, existing benchmarks for moral and clinical decision-making do not adequately capture the unique ethical dilemmas encountered in mental health practice, where confidentiality, autonomy, beneficence, and bias frequently intersect. To address this gap, we introduce Ethical Reasoning in Mental Health (EthicsMH), a pilo…

Mendelian Randomization Methods for Causal Inference: Estimands, Identification and Inference
Minhao Yao, Anqi Wang, Xihao Li, Zhonghua Liu
https://arxiv.org/abs/2509.11519 http…

Mendelian Randomization Methods for Causal Inference: Estimands, Identification and Inference
Mendelian randomization (MR) has become an essential tool for causal inference in biomedical and public health research. By using genetic variants as instrumental variables, MR helps address unmeasured confounding and reverse causation, offering a quasi-experimental framework to evaluate causal effects of modifiable exposures on health outcomes. Despite its promise, MR faces substantial methodological challenges, including invalid instruments, weak instrument bias, and design complexities acros…

Addressing Bias in VLMs for Glaucoma Detection Without Protected Attribute Supervision
Ahsan Habib Akash, Greg Murray, Annahita Amireskandari, Joel Palko, Carol Laxson, Binod Bhattarai, Prashnna Gyawali
https://arxiv.org/abs/2508.09087

Addressing Bias in VLMs for Glaucoma Detection Without Protected Attribute Supervision
Vision-Language Models (VLMs) have achieved remarkable success on multimodal tasks such as image-text retrieval and zero-shot classification, yet they can exhibit demographic biases even when explicit protected attributes are absent during training. In this work, we focus on automated glaucoma screening from retinal fundus images, a critical application given that glaucoma is a leading cause of irreversible blindness and disproportionately affects underserved populations. Building on a reweight…

SCDF: A Speaker Characteristics DeepFake Speech Dataset for Bias Analysis
Vojt\v{e}ch Stan\v{e}k, Karel Srna, Anton Firc, Kamil Malinka
https://arxiv.org/abs/2508.07944 https://…

SCDF: A Speaker Characteristics DeepFake Speech Dataset for Bias Analysis
Despite growing attention to deepfake speech detection, the aspects of bias and fairness remain underexplored in the speech domain. To address this gap, we introduce the Speaker Characteristics Deepfake (SCDF) dataset: a novel, richly annotated resource enabling systematic evaluation of demographic biases in deepfake speech detection. SCDF contains over 237,000 utterances in a balanced representation of both male and female speakers spanning five languages and a wide age range. We evaluate seve…

On Inherited Popularity Bias in Cold-Start Item Recommendation
Gregor Meehan, Johan Pauwels
https://arxiv.org/abs/2510.11402 https://arxiv.org/pdf/2510.114…

On Inherited Popularity Bias in Cold-Start Item Recommendation
Collaborative filtering (CF) recommender systems struggle with making predictions on unseen, or 'cold', items. Systems designed to address this challenge are often trained with supervision from warm CF models in order to leverage collaborative and content information from the available interaction data. However, since they learn to replicate the behavior of CF methods, cold-start models may therefore also learn to imitate their predictive biases. In this paper, we show that cold-start systems c…

Pangenome-guided sequence assembly via binary optimisation
Josh Cudby, James Bonfield, Chenxi Zhou, Richard Durbin, Sergii Strelchuk
https://arxiv.org/abs/2508.08200 https://

Pangenome-guided sequence assembly via binary optimisation
De novo genome assembly is challenging in highly repetitive regions; however, reference-guided assemblers often suffer from bias. We propose a framework for pangenome-guided sequence assembly, which can resolve short-read data in complex regions without bias towards a single reference genome. Our method frames assembly as a graph traversal optimisation problem, which can be implemented on quantum computers. The pipeline first annotates pangenome graphs with estimated copy numbers for each node,…

Query-Focused Extractive Summarization for Sentiment Explanation
Ahmed Moubtahij, Sylvie Ratt\'e, Yazid Attabi, Maxime Dumas
https://arxiv.org/abs/2509.11989 https://…

Query-Focused Extractive Summarization for Sentiment Explanation
Constructive analysis of feedback from clients often requires determining the cause of their sentiment from a substantial amount of text documents. To assist and improve the productivity of such endeavors, we leverage the task of Query-Focused Summarization (QFS). Models of this task are often impeded by the linguistic dissonance between the query and the source documents. We propose and substantiate a multi-bias framework to help bridge this gap at a domain-agnostic, generic level; we then for…

BSI veröffentlicht Whitepaper zum Bias in der KI
Wenn sich KI-Systeme nicht so verhalten, wie erwünscht, stecken oft Verzerrungen in den Daten dahinter. Das BSI erklärt, wie man den Bias im System erkennt.
https://www.

Not in Sync: Unveiling Temporal Bias in Audio Chat Models
Jiayu Yao, Shenghua Liu, Yiwei Wang, Rundong Cheng, Lingrui Mei, Baolong Bi, Zhen Xiong, Xueqi Cheng
https://arxiv.org/abs/2510.12185

Not in Sync: Unveiling Temporal Bias in Audio Chat Models
Large Audio Language Models (LALMs) are increasingly applied to audio understanding and multimodal reasoning, yet their ability to locate when events occur remains underexplored. We present the first systematic study of temporal bias in LALMs, revealing a key limitation in their timestamp prediction. For example, when asked "At which second does the lecturer introduce the key formula?", models often predict timestamps that are consistently earlier or later than the ground truth. Through control…

Dr. Bias: Social Disparities in AI-Powered Medical Guidance
Emma Kondrup, Anne Imouza
https://arxiv.org/abs/2510.09162 https://arxiv.org/pdf/2510.09162

Dr. Bias: Social Disparities in AI-Powered Medical Guidance
With the rapid progress of Large Language Models (LLMs), the general public now has easy and affordable access to applications capable of answering most health-related questions in a personalized manner. These LLMs are increasingly proving to be competitive, and now even surpass professionals in some medical capabilities. They hold particular promise in low-resource settings, considering they provide the possibility of widely accessible, quasi-free healthcare support. However, evaluations that …

Discrimination by LLMs: Cross-lingual Bias Assessment and Mitigation in Decision-Making and Summarisation
Willem Huijzer, Jieying Chen
https://arxiv.org/abs/2509.09735 https://

Discrimination by LLMs: Cross-lingual Bias Assessment and Mitigation in Decision-Making and Summarisation
The rapid integration of Large Language Models (LLMs) into various domains raises concerns about societal inequalities and information bias. This study examines biases in LLMs related to background, gender, and age, with a focus on their impact on decision-making and summarization tasks. Additionally, the research examines the cross-lingual propagation of these biases and evaluates the effectiveness of prompt-instructed mitigation strategies. Using an adapted version of the dataset by Tamkin et…

OpenAI says GPT‑5 instant and GPT‑5 thinking cut political bias by 30% from earlier models, and show greater robustness to charged prompts (Ashley Gold/Axios)
https://www.axios.com/2025/10/09/openai-gpt-5-least-biased-model

OpenAI says GPT-5 is its least biased model yet
New findings from OpenAI researchers found that "charged" prompts produced the most biased results.

Does LLM Focus on the Right Words? Diagnosing Language Bias in LLM-based Recommenders
Bohao Wang, Jiawei Chen, Feng Liu, Changwang Zhang, Jun Wang, Canghong Jin, Chun Chen, Can Wang
https://arxiv.org/abs/2510.10978

Does LLM Focus on the Right Words? Diagnosing Language Bias in LLM-based Recommenders
Large language models (LLMs), owing to their extensive open-domain knowledge and semantic reasoning capabilities, have been increasingly integrated into recommender systems (RS). However, a substantial gap remains between the pre-training objectives of LLMs and the specific requirements of recommendation tasks. To address this gap, supervised fine-tuning (SFT) is commonly performed on specially curated recommendation datasets to further enhance their predictive ability. Despite its success, SFT…

Chronologically Consistent Generative AI
Songrun He, Linying Lv, Asaf Manela, Jimmy Wu
https://arxiv.org/abs/2510.11677 https://arxiv.org/pdf/2510.11677

Chronologically Consistent Generative AI
We introduce a family of chronologically consistent, instruction-following large language models to eliminate lookahead bias. Each model is trained only on data available before a clearly defined knowledge-cutoff date, ensuring strict temporal separation from any post-cutoff data. The resulting framework offers (i) a simple, conversational chat interface, (ii) fully open, fixed model weights that guarantee replicability, and (iii) a conservative lower bound on forecast accuracy, isolating the s…

Yet another algorithmic bias: A Discursive Analysis of Large Language Models Reinforcing Dominant Discourses on Gender and Race
Gustavo Bonil, Simone Hashiguti, Jhessica Silva, Jo\~ao Gondim, Helena Maia, N\'adia Silva, Helio Pedrini, Sandra Avila
https://arxiv.org/abs/2508.10304

Yet another algorithmic bias: A Discursive Analysis of Large Language Models Reinforcing Dominant Discourses on Gender and Race
With the advance of Artificial Intelligence (AI), Large Language Models (LLMs) have gained prominence and been applied in diverse contexts. As they evolve into more sophisticated versions, it is essential to assess whether they reproduce biases, such as discrimination and racialization, while maintaining hegemonic discourses. Current bias detection approaches rely mostly on quantitative, automated methods, which often overlook the nuanced ways in which biases emerge in natural language. This st…

PakBBQ: A Culturally Adapted Bias Benchmark for QA
Abdullah Hashmat, Muhammad Arham Mirza, Agha Ali Raza
https://arxiv.org/abs/2508.10186 https://arxiv.org…

PakBBQ: A Culturally Adapted Bias Benchmark for QA
With the widespread adoption of Large Language Models (LLMs) across various applications, it is empirical to ensure their fairness across all user communities. However, most LLMs are trained and evaluated on Western centric data, with little attention paid to low-resource languages and regional contexts. To address this gap, we introduce PakBBQ, a culturally and regionally adapted extension of the original Bias Benchmark for Question Answering (BBQ) dataset. PakBBQ comprises over 214 templates,…

Causal Inspired Multi Modal Recommendation
Jie Yang, Chenyang Gu, Zixuan Liu
https://arxiv.org/abs/2510.12325 https://arxiv.org/pdf/2510.12325

Causal Inspired Multi Modal Recommendation
Multimodal recommender systems enhance personalized recommendations in e-commerce and online advertising by integrating visual, textual, and user-item interaction data. However, existing methods often overlook two critical biases: (i) modal confounding, where latent factors (e.g., brand style or product category) simultaneously drive multiple modalities and influence user preference, leading to spurious feature-preference associations; (ii) interaction bias, where genuine user preferences are m…

KI-Update: GPT-5, Google-Shopping, Bias in der KI, Unitree R1
Das "KI-Update" liefert werktäglich eine Zusammenfassung der wichtigsten KI-Entwicklungen.
https://www.heise.de…

Knowing Unknowns in an Age of Information Overload
Saurabh Khanna
https://arxiv.org/abs/2510.10413 https://arxiv.org/pdf/2510.10413

Knowing Unknowns in an Age of Information Overload
The technological revolution of the Internet has digitized the social, economic, political, and cultural activities of billions of humans. While researchers have been paying due attention to concerns of misinformation and bias, these obscure a much less researched and equally insidious problem - that of uncritically consuming incomplete information. The problem of incomplete information consumption stems from the very nature of explicitly ranked information on digital platforms, where our limit…

An Automated Multi-Modal Evaluation Framework for Mobile Intelligent Assistants
Meiping Wang, Jian Zhong, Rongduo Han, Liming Kang, Zhengkun Shi, Xiao Liang, Xing Lin, Nan Gao, Haining Zhang
https://arxiv.org/abs/2508.09507

An Automated Multi-Modal Evaluation Framework for Mobile Intelligent Assistants
With the rapid development of mobile intelligent assistant technologies, multi-modal AI assistants have become essential interfaces for daily user interactions. However, current evaluation methods face challenges including high manual costs, inconsistent standards, and subjective bias. This paper proposes an automated multi-modal evaluation framework based on large language models and multi-agent collaboration. The framework employs a three-tier agent architecture consisting of interaction eval…

A Research Vision for Web Search on Emerging Topics
Alisa Rieger, Stefan Dietze, Ran Yu
https://arxiv.org/abs/2509.10212 https://arxiv.org/pdf/2509.10212…

A Research Vision for Web Search on Emerging Topics
We regularly encounter information on novel, emerging topics for which the body of knowledge is still evolving, which can be linked, for instance, to current events. A primary way to learn more about such topics is through web search. However, information on emerging topics is sparse and evolves dynamically as knowledge grows, making it uncertain and variable in quality and trustworthiness and prone to deliberate or accidental manipulation, misinformation, and bias. In this paper, we outline a …

Tokenization Disparities as Infrastructure Bias: How Subword Systems Create Inequities in LLM Access and Efficiency
Hailay Kidu Teklehaymanot, Wolfgang Nejdl
https://arxiv.org/abs/2510.12389

Tokenization Disparities as Infrastructure Bias: How Subword Systems Create Inequities in LLM Access and Efficiency
Tokenization disparities pose a significant barrier to achieving equitable access to artificial intelligence across linguistically diverse populations. This study conducts a large-scale cross-linguistic evaluation of tokenization efficiency in over 200 languages to systematically quantify computational inequities in large language models (LLMs). Using a standardized experimental framework, we applied consistent preprocessing and normalization protocols, followed by uniform tokenization through …

The Hidden Bias: A Study on Explicit and Implicit Political Stereotypes in Large Language Models
Konrad L\"ohr, Shuzhou Yuan, Michael F\"arber
https://arxiv.org/abs/2510.08236

The Hidden Bias: A Study on Explicit and Implicit Political Stereotypes in Large Language Models
Large Language Models (LLMs) are increas- ingly integral to information dissemination and decision-making processes. Given their grow- ing societal influence, understanding potential biases, particularly within the political domain, is crucial to prevent undue influence on public opinion and democratic processes. This work investigates political bias and stereotype propa- gation across eight prominent LLMs using the two-dimensional Political Compass Test (PCT). Initially, the PCT is employed to…

Examining the Association between Estimated Prevalence and Diagnostic Test Accuracy using Directed Acyclic Graphs
Yang Lu, Robert Platt, Nandini Dendukuri
https://arxiv.org/abs/2508.10207

Examining the Association between Estimated Prevalence and Diagnostic Test Accuracy using Directed Acyclic Graphs
There have been reports of correlation between estimates of prevalence and test accuracy across studies included in diagnostic meta-analyses. It has been hypothesized that this unexpected association arises because of certain biases commonly found in diagnostic accuracy studies. A theoretical explanation has not been studied systematically. In this work, we introduce directed acyclic graphs to illustrate common structures of bias in diagnostic test accuracy studies and to define the resulting d…

Detecting Gender Stereotypes in Scratch Programming Tutorials
Isabella Gra{\ss}l, Benedikt Fein, Gordon Fraser
https://arxiv.org/abs/2510.11064 https://arx…

Detecting Gender Stereotypes in Scratch Programming Tutorials
Gender stereotypes in introductory programming courses often go unnoticed, yet they can negatively influence young learners' interest and learning, particularly under-represented groups such as girls. Popular tutorials on block-based programming with Scratch may unintentionally reinforce biases through character choices, narrative framing, or activity types. Educators currently lack support in identifying and addressing such bias. With large language models~(LLMs) increasingly used to generate …

The Fair Game: Auditing & Debiasing AI Algorithms Over Time
Debabrota Basu, Udvas Das
https://arxiv.org/abs/2508.06443 https://arxiv.org/pdf/2508.06443…

The Fair Game: Auditing & Debiasing AI Algorithms Over Time
An emerging field of AI, namely Fair Machine Learning (ML), aims to quantify different types of bias (also known as unfairness) exhibited in the predictions of ML algorithms, and to design new algorithms to mitigate them. Often, the definitions of bias used in the literature are observational, i.e. they use the input and output of a pre-trained algorithm to quantify a bias under concern. In reality,these definitions are often conflicting in nature and can only be deployed if either the ground t…

Mitigating Popularity Bias in Counterfactual Explanations using Large Language Models
Arjan Hasami, Masoud Mansoury
https://arxiv.org/abs/2508.08946 https://

Mitigating Popularity Bias in Counterfactual Explanations using Large Language Models
Counterfactual explanations (CFEs) offer a tangible and actionable way to explain recommendations by showing users a "what-if" scenario that demonstrates how small changes in their history would alter the system's output. However, existing CFE methods are susceptible to bias, generating explanations that might misalign with the user's actual preferences. In this paper, we propose a pre-processing step that leverages large language models to filter out-of-character history items before generatin…

HALF: Harm-Aware LLM Fairness Evaluation Aligned with Deployment
Ali Mekky, Omar El Herraoui, Preslav Nakov, Yuxia Wang
https://arxiv.org/abs/2510.12217 https://

HALF: Harm-Aware LLM Fairness Evaluation Aligned with Deployment
Large language models (LLMs) are increasingly deployed across high-impact domains, from clinical decision support and legal analysis to hiring and education, making fairness and bias evaluation before deployment critical. However, existing evaluations lack grounding in real-world scenarios and do not account for differences in harm severity, e.g., a biased decision in surgery should not be weighed the same as a stylistic bias in text summarization. To address this gap, we introduce HALF (Harm-A…

Bias correction for Chatterjee's graph-based correlation coefficient
Mona Azadkia, Leihao Chen, Fang Han
https://arxiv.org/abs/2508.09040 https://arxiv…

Bias correction for Chatterjee's graph-based correlation coefficient
Azadkia and Chatterjee (2021) recently introduced a simple nearest neighbor (NN) graph-based correlation coefficient that consistently detects both independence and functional dependence. Specifically, it approximates a measure of dependence that equals 0 if and only if the variables are independent, and 1 if and only if they are functionally dependent. However, this NN estimator includes a bias term that may vanish at a rate slower than root-$n$, preventing root-$n$ consistency in general. In …

Bridging the Physics-Data Gap with FNO-Guided Conditional Flow Matching: Designing Inductive Bias through Hierarchical Physical Constraints
Tsuyoshi Okita
https://arxiv.org/abs/2510.08295

Bridging the Physics-Data Gap with FNO-Guided Conditional Flow Matching: Designing Inductive Bias through Hierarchical Physical Constraints
Conventional time-series generation often ignores domain-specific physical constraints, limiting statistical and physical consistency. We propose a hierarchical framework that embeds the inherent hierarchy of physical laws-conservation, dynamics, boundary, and empirical relations-directly into deep generative models, introducing a new paradigm of physics-informed inductive bias. Our method combines Fourier Neural Operators (FNOs) for learning physical operators with Conditional Flow Matching (C…

Evaluating and comparing gender bias across four text-to-image models
Zoya Hammad, Nii Longdon Sowah
https://arxiv.org/abs/2509.08004 https://arxiv.org/pdf…

Evaluating and comparing gender bias across four text-to-image models
As we increasingly use Artificial Intelligence (AI) in decision-making for industries like healthcare, finance, e-commerce, and even entertainment, it is crucial to also reflect on the ethical aspects of AI, for example the inclusivity and fairness of the information it provides. In this work, we aimed to evaluate different text-to-image AI models and compare the degree of gender bias they present. The evaluated models were Stable Diffusion XL (SDXL), Stable Diffusion Cascade (SC), DALL-E and E…

SBS: Enhancing Parameter-Efficiency of Neural Representations for Neural Networks via Spectral Bias Suppression
Qihu Xie, Yuan Li, Yi Kang
https://arxiv.org/abs/2509.07373 https…

SBS: Enhancing Parameter-Efficiency of Neural Representations for Neural Networks via Spectral Bias Suppression
Implicit neural representations have recently been extended to represent convolutional neural network weights via neural representation for neural networks, offering promising parameter compression benefits. However, standard multi-layer perceptrons used in neural representation for neural networks exhibit a pronounced spectral bias, hampering their ability to reconstruct high-frequency details effectively. In this paper, we propose SBS, a parameter-efficient enhancement to neural representatio…

Z-Curve Plot: A Visual Diagnostic for Publication Bias in Meta-Analysis
Franti\v{s}ek Barto\v{s}, Ulrich Schimmack
https://arxiv.org/abs/2509.07171 https://

Z-Curve Plot: A Visual Diagnostic for Publication Bias in Meta-Analysis
Publication bias undermines meta-analytic inference, yet visual diagnostics for detecting model misfit due to publication bias are lacking. We propose the z-curve plot, a publication-bias-focused absolute model fit diagnostic. The z-curve plot overlays the model-implied posterior predictive distribution of $z$-statistics on the observed distribution of z-statistics and enables direct comparison across candidate meta-analytic models (e.g., random-effects meta-analysis, selection models, PET-PEES…

Fisher Information, Training and Bias in Fourier Regression Models
Lorenzo Pastori, Veronika Eyring, Mierk Schwabe
https://arxiv.org/abs/2510.06945 https://

Fisher Information, Training and Bias in Fourier Regression Models
Motivated by the growing interest in quantum machine learning, in particular quantum neural networks (QNNs), we study how recently introduced evaluation metrics based on the Fisher information matrix (FIM) are effective for predicting their training and prediction performance. We exploit the equivalence between a broad class of QNNs and Fourier models, and study the interplay between the \emph{effective dimension} and the \emph{bias} of a model towards a given task, investigating how these affe…

Echoes of Agreement: Argument Driven Opinion Shifts in Large Language Models
Avneet Kaur
https://arxiv.org/abs/2508.09759 https://arxiv.org/pdf/2508.09759

Echoes of Agreement: Argument Driven Opinion Shifts in Large Language Models
There have been numerous studies evaluating bias of LLMs towards political topics. However, how positions towards these topics in model outputs are highly sensitive to the prompt. What happens when the prompt itself is suggestive of certain arguments towards those positions remains underexplored. This is crucial for understanding how robust these bias evaluations are and for understanding model behaviour, as these models frequently interact with opinionated text. To that end, we conduct experim…

Acquiescence Bias in Large Language Models
Daniel Braun
https://arxiv.org/abs/2509.08480 https://arxiv.org/pdf/2509.08480

Acquiescence Bias in Large Language Models
Acquiescence bias, i.e. the tendency of humans to agree with statements in surveys, independent of their actual beliefs, is well researched and documented. Since Large Language Models (LLMs) have been shown to be very influenceable by relatively small changes in input and are trained on human-generated data, it is reasonable to assume that they could show a similar tendency. We present a study investigating the presence of acquiescence bias in LLMs across different models, tasks, and languages …

From Detection to Mitigation: Addressing Gender Bias in Chinese Texts via Efficient Tuning and Voting-Based Rebalancing
Chengyan Wu, Yiqiang Cai, Yufei Cheng, Yun Xue
https://arxiv.org/abs/2509.07889

From Detection to Mitigation: Addressing Gender Bias in Chinese Texts via Efficient Tuning and Voting-Based Rebalancing
This paper presents our team's solution to Shared Task 7 of NLPCC-2025, which focuses on sentence-level gender bias detection and mitigation in Chinese. The task aims to promote fairness and controllability in natural language generation by automatically detecting, classifying, and mitigating gender bias. To address this challenge, we adopt a fine-tuning approach based on large language models (LLMs), efficiently adapt to the bias detection task via Low-Rank Adaptation (LoRA). In terms of data …

GAMBIT : A Challenge Set for Evaluating Gender Bias in Machine Translation Quality Estimation Metrics
Giorgos Filandrianos, Orfeas Menis Mastromichalakis, Wafaa Mohammed, Giuseppe Attanasio, Chrysoula Zerva
https://arxiv.org/abs/2510.06841

GAMBIT+: A Challenge Set for Evaluating Gender Bias in Machine Translation Quality Estimation Metrics
Gender bias in machine translation (MT) systems has been extensively documented, but bias in automatic quality estimation (QE) metrics remains comparatively underexplored. Existing studies suggest that QE metrics can also exhibit gender bias, yet most analyses are limited by small datasets, narrow occupational coverage, and restricted language variety. To address this gap, we introduce a large-scale challenge set specifically designed to probe the behavior of QE metrics when evaluating translat…

Bias after Prompting: Persistent Discrimination in Large Language Models
Nivedha Sivakumar, Natalie Mackraz, Samira Khorshidi, Krishna Patel, Barry-John Theobald, Luca Zappella, Nicholas Apostoloff
https://arxiv.org/abs/2509.08146

Bias after Prompting: Persistent Discrimination in Large Language Models
A dangerous assumption that can be made from prior work on the bias transfer hypothesis (BTH) is that biases do not transfer from pre-trained large language models (LLMs) to adapted models. We invalidate this assumption by studying the BTH in causal models under prompt adaptations, as prompting is an extremely popular and accessible adaptation strategy used in real-world applications. In contrast to prior work, we find that biases can transfer through prompting and that popular prompt-based mit…

Mitigating Judgment Preference Bias in Large Language Models through Group-Based Polling
Shuliang Liu, Zhipeng Xu, Zhenghao Liu, Yukun Yan, Minghe Yu, Yu Gu, Chong Chen, Huiyuan Xie, Ge Yu
https://arxiv.org/abs/2510.08145

Mitigating Judgment Preference Bias in Large Language Models through Group-Based Polling
Large Language Models (LLMs) as automatic evaluators, commonly referred to as LLM-as-a-Judge, have also attracted growing attention. This approach plays a vital role in aligning LLMs with human judgments, providing accurate and reliable assessments. However, LLM-based judgment models often exhibit judgment preference bias during the evaluation phase, tending to favor responses generated by themselves, undermining the reliability of their judgments. This paper introduces the Group-Based Polling …

Measuring Bias or Measuring the Task: Understanding the Brittle Nature of LLM Gender Biases
Bufan Gao, Elisa Kreiss
https://arxiv.org/abs/2509.04373 https://

Measuring Bias or Measuring the Task: Understanding the Brittle Nature of LLM Gender Biases
As LLMs are increasingly applied in socially impactful settings, concerns about gender bias have prompted growing efforts both to measure and mitigate such bias. These efforts often rely on evaluation tasks that differ from natural language distributions, as they typically involve carefully constructed task prompts that overtly or covertly signal the presence of gender bias-related content. In this paper, we examine how signaling the evaluative purpose of a task impacts measured gender bias in …

Simulating Identity, Propagating Bias: Abstraction and Stereotypes in LLM-Generated Text
Pia Sommerauer, Giulia Rambelli, Tommaso Caselli
https://arxiv.org/abs/2509.08484 https:…

Simulating Identity, Propagating Bias: Abstraction and Stereotypes in LLM-Generated Text
Persona-prompting is a growing strategy to steer LLMs toward simulating particular perspectives or linguistic styles through the lens of a specified identity. While this method is often used to personalize outputs, its impact on how LLMs represent social groups remains underexplored. In this paper, we investigate whether persona-prompting leads to different levels of linguistic abstraction - an established marker of stereotyping - when generating short texts linking socio-demographic categories…