Unintended Consequences of Recommender System Interventions: Evidence from a Field Experiment

2026-06-06T17:21:22Z

Platform content interventions in recommendation systems are typically evaluated as static "nudges", ignoring that the systems adaptively learn from the resulting user behavior. We investigate this dynamic through a large-scale field experiment on a short-video platform. The experiment involves a "sleep reminder" campaign designed to reduce late-night usage. Paradoxically, the intervention increased late-night engagement by 14.75% and overall platform usage by 2.18%, and the effects persisted for weeks even after the experiment. We explain this through a forced-exploration mechanism, showing that by revealing high latent demand for the promoted content, the intervention triggers a recommendation policy update that routine user behavior would not produce. The data generated by the intervention induced the algorithm to update its post-campaign policy, reinforcing the very engagement loops the campaign aimed to mitigate. Our findings demonstrate that user-facing interventions can effectively retrain the underlying algorithm, triggering durable, system-wide shifts in content distribution that challenge standard evaluation metrics in platform governance and social responsibility initiatives.

AI Self-preferencing in Algorithmic Hiring: Empirical Evidence and Insights

2026-06-06T15:26:51Z

As artificial intelligence (AI) tools become widely adopted, large language models (LLMs) are increasingly involved on both sides of decision-making processes, ranging from hiring to content moderation. This dual adoption raises a critical question: do LLMs systematically favor content that resembles their own outputs? Prior research in computer science has identified self-preference bias -- the tendency of LLMs to favor their own generated content -- but its real-world implications have not been empirically evaluated. We focus on the hiring context, where job applicants often rely on LLMs to refine resumes, while employers deploy them to screen those same resumes. Using a large-scale controlled resume correspondence experiment, we find that LLMs consistently prefer resumes generated by themselves over those written by humans or produced by alternative models, even when content quality is controlled. The bias against human-written resumes is particularly substantial, with self-preference bias ranging from 67% to 82% across major commercial and open-source models. To assess labor market impact, we simulate realistic hiring pipelines across 24 occupations. These simulations show that candidates using the same LLM as the evaluator are 23% to 60% more likely to be shortlisted than equally qualified applicants submitting human-written resumes, with the largest disadvantages observed in business-related fields such as sales and accounting. We further demonstrate that this bias can be reduced by more than 50% through simple interventions targeting LLMs' self-recognition capabilities. These findings highlight an emerging but previously overlooked risk in AI-assisted decision making and call for expanded frameworks of AI fairness that address not only demographic-based disparities, but also biases in AI-AI interactions.

The Governance of Human-LLM Interaction: Safety Gating, Civility Steering, and Affective Default Lock-In

2026-06-06T13:36:37Z

Large language models (LLMs) increasingly mediate high-stakes interactions in finance, medicine, and mental-health support, yet users have limited control over how these systems communicate. We frame interaction style as a governance object: provider-side alignment not only blocks harmful content, but also stabilizes communicative defaults that shape users' epistemic distance, relational expectations, and capacity to opt out of emotionalized or anthropomorphic interaction. We introduce a deterministic multi-agent evaluation pipeline for measuring prompt steerability and style drift in long-horizon dialogue. The study replays 100 frozen user-only scripts across four domains and three runnable persona conditions: default, sarcastic, and cold, using three generator models, yielding 90,000 assistant replies scored by a human-calibrated LLM judge on harmfulness, negative emotion, inappropriateness, empathic language, anthropomorphism, and refusal behavior. A fourth harmful persona is evaluated separately as a safety-gating test. The paper contributes a reproducible method for quantifying whether prompt-specified styles remain stable over time and a governance framework distinguishing safety gating, civility steering, and affective default lock-in. Overall, we show that prompt steerability and regression-to-default are observable indicators of provider control over communicative form, with implications for pluralism, autonomy, and democratic agency in human-LLM interaction.

XCR-Bench: Benchmarking Cross-Cultural Reasoning in LLMs via Culture-Specific Items and Hall's Triad

2026-06-06T11:19:39Z

Cross-cultural competence in large language models (LLMs) requires understanding and adapting Culture-Specific Items (CSIs) across varying cultural contexts. However, progress in evaluating this capability remains limited by the lack of high-quality CSI-annotated corpora with parallel cross-cultural sentence pairs. We introduce XCR-Bench, a Cross(X)-Cultural Reasoning Benchmark containing 4.1k parallel sentences and 1,098 CSIs across three reasoning tasks. XCR-Bench integrates Newmark's CSI framework with Hall's Triad of Culture, enabling evaluation across levels of cultural visibility -- from observable practices to implicit social norms and values. Experiments on eight multilingual LLMs show that state-of-the-art models exhibit consistent weaknesses in identifying and adapting specific categories of CSIs, revealing a gap between surface-level recall and explicit cultural reasoning. Performance declines significantly on culturally sensitive categories and deeper cultural levels (p<0.005, 8/8 models), and adaptation quality varies systematically across target cultures and Bengali regional variants, indicating encoded regional and ethno-religious biases even within a single linguistic setting. We publicly release the corpus and code to support future research on cross-cultural NLP.

Data Want to be Free: An Innovation Resistance Theory Model for Identifying Barriers to Government Data Sharing

2026-06-06T10:05:07Z

Data sharing is increasingly essential for digital government and data-driven innovation, yet many public organizations remain reluctant to make their data openly available. While prior research has examined factors influencing open data adoption, little theoretical work explores why resistance persists within public agencies. This study develops an Innovation Resistance Theory (IRT) model tailored to government data sharing to identify predictors of organizational resistance. An initial model was derived from literature and refined through interviews with 21 public organizations across six European countries. The resulting IRT4DS model identifies 39 barriers spanning usage, value, risk, tradition, and image dimensions, and 23 countermeasures mapped to the most critical barriers and the actors responsible for addressing them. By extending IRT into the context of governmental data sharing, the study advances theoretical understanding of why public data often remains closed and provides actionable guidance for policymakers seeking to design enabling data ecosystems and reduce structural and cultural barriers to OGD adoption.

"I understand your perspective": LLM Persuasion and Sycophancy through the Lens of Communicative Action Theory

2026-06-06T09:54:31Z

Large Language Models (LLMs) can generate high-quality arguments, yet their ability to engage in nuanced and persuasive communicative actions remains largely unexplored. This work explores the persuasive potential of LLMs through the framework of Jürgen Habermas' Theory of Communicative Action. It examines whether LLMs express illocutionary intent (i.e., pragmatic functions of language such as conveying knowledge, building trust, or signaling similarity) in ways that are comparable to human communication. We simulate online discussions between opinion holders and LLMs using conversations from the persuasive subreddit ChangeMyView. We then compare the likelihood of illocutionary intents in human-written and LLM-generated counter-arguments, specifically those that successfully changed the original poster's view. We find that all three LLMs effectively convey illocutionary intent -- often more so than humans -- potentially increasing their anthropomorphism. Further, LLMs craft sycophantic responses that closely align with the opinion holder's intent, a strategy strongly associated with opinion change. Finally, crowd-sourced workers find LLM-generated counter-arguments more agreeable and consistently prefer them over human-written ones. These findings suggest that LLMs' persuasive power extends beyond merely generating high-quality arguments. On the contrary, training LLMs with human preferences effectively tunes them to mirror human communication patterns, particularly nuanced communicative actions, potentially increasing individuals' susceptibility to their influence.

Predictive Volatility of Machine Learning in Micro-Samples: A Regularised Assessment of Regional Poverty

2026-06-06T07:22:29Z

Small regional datasets pose a dual statistical problem: correlated predictors inflate estimation variance, while flexible learners can become unstable because the available information per adaptive degree of freedom is limited. We examine this issue through predictive volatility, defined as the cross-sample dispersion and upper-tail behaviour of out-of-sample loss. Using simulation evidence reported for sparse linear, near-linear and heavy-tailed settings, we compare ordinary least squares, frequentist penalties, Bayesian shrinkage models, bounded-response and spatial specifications, and flexible machine-learning procedures. In the reported simulation results, regularised linear estimators generally dominate in the linear high-collinearity micro-sample settings and remain the most reliable overall, whereas tree-based methods become more competitive only when the signal is weakly nonlinear and the sample size is larger. In the empirical application to 34 Indonesian provinces, ridge yields the best leave-one-out performance, followed by elastic net and lasso. Across the Bayesian shrinkage specifications, ICT skills show the most consistent negative association with poverty, with the strongest support under horseshoe and spike-and-slab formulations. These results suggest that, in micro-sample regional modelling, the main constraint is limited information per effective degree of freedom rather than insufficient algorithmic flexibility.

Cognitive Comparability and the Limits of Governance: Evaluating Authority Under Radical Capability Asymmetry

2026-06-06T04:29:46Z

Governance theory presupposes a rough cognitive comparability between governors and governed. This paper makes that assumption explicit and testable through a six-dimension evaluation framework covering legitimacy, accountability, corrigibility, non-domination, subsidiarity, and institutional resilience, drawn from political legitimacy theory, principal-agent models, republican theory, and the AI alignment literature. The framework is first demonstrated on existing non-majoritarian institutions, where capability asymmetry is real but bounded, and then applied to a prospective case of bounded superintelligent authority, where the asymmetry is radical. Four of six dimensions show structural failures. Two of the four appear tractable to institutional design (subsidiarity scope limitation and institutional resilience). The other two, the public reason problem under cognitive incomprehensibility and the non-domination problem under permanent capability asymmetry, call for new normative theory rather than better institutional design. The analysis also finds that dimensions which operate as independent checks under bounded asymmetry begin to degrade together under radical asymmetry, because each depends on the same oversight capacity. The assumptions that allowed these checks to remain independent have gone unexamined so far because they have always held.

EduMirror: Modeling Educational Social Dynamics with Value-driven Multi-agent Simulation

2026-06-06T02:38:30Z

Understanding how educational social dynamics evolve is critical for informing effective educational policies and counterfactual interventions. However, traditional methods face a fundamental dilemma: observational studies often lack causal power, while controlled experiments are frequently constrained by ethical concerns. Although LLM-based multi-agent simulations offer a scalable in silico alternative, existing approaches remain limited by weak psychological grounding and insufficient measurement of latent psychological states. To address this, we introduce EduMirror, a multi-agent simulator for the scientific study of educational social dynamics. We provide configurable education-oriented agent forms, including value-driven agents grounded in psychological needs and social value orientation, together with a dual-track measurement protocol for quantifying observable behaviors and latent psychological states. We validate the realism and usability of EduMirror through case studies on school bullying and group cooperation, as well as broader evaluations across diverse educational scenarios. The results show that EduMirror generates educational social dynamics that are realistic, theory-consistent, and measurable by empirical criteria. These properties enable structured in silico educational research, providing a computational tool for hypothesis testing and counterfactual intervention analysis in educational science. Project page: https://edumirror.net.

Stable Geometry, Reversing Poles: The Bipolar Structure of AI Occupational Substitutability and Its Decade-Scale Inversion

2026-06-06T02:00:32Z

Empirical research on the labor-market impact of artificial intelligence has converged, since Frey and Osborne (2017), on a continuous-gradient representation in which each occupation is assigned a real-valued exposure score on [0,1] obtained by linear aggregation across capability dimensions. This continuity is rarely articulated as an assumption and has not been tested at the micro-action level where substitution actually occurs. We decompose 1,961 O*NET Detailed Work Activities into 15,817 micro-actions using a multi-agent LLM pipeline with 31-expert HITL calibration, then project the DWA-level Occupational Automation Index from our prior work onto a 7-macro semantic typology. The result is a bipolar structure. Tool-Mediated Physical (M2, mean OAI = 0.054) and Planning & Design (M7, mean OAI = 0.499) form two extremes separated by Cohen's d = 2.41 (H = 172.88, p = 6.21e-34). The geometry is robust under three independent stress tests: resolution (K=7 to K=15, polar gap widens from 0.45 to 0.57), encoder swap to BGE (LLM-class OAI lead replicates at 3.37x), and Eloundou's GPT-4 task ratings (DWA-level rho = 0.635). The six middle macros form a low-contrast band between the poles (TOST at d=0.2 admits only 1/15 pairs as equivalent), not a flat plain. The geometry's stability does not, however, extend to its content. Across a decade, the polarity has inverted. Frey-Osborne (2013) placed Tool-Mediated Physical near the highest computerisation risk and Planning & Design near the lowest; our LLM-era OAI reverses that order, with macro-level FO-Eloundou Spearman rho = -0.750, p = 0.020, against the original Oxford Martin appendix. Which pole is high is therefore contingent on the era's dominant capability frontier, while the stable geometry itself is the structurally robust object.

Learning Behavioral Signals from Encrypted Smartphone Network Traffic

2026-06-05T22:55:35Z

Human behavior is challenging to measure continuously at scale, yet traces of daily routines and well-being may be reflected in interactions with personal devices. We investigate whether encrypted smartphone network traffic can serve as a passive sensing signal for behavioral states related to sleep disturbance, stress, and loneliness. To capture both population-level patterns and individual-specific behavior, we employ a transformer-based model with user-specific adapters that learns representations of network activity while accounting for personal baselines and deviations from them. To improve interpretability, we further analyze these representations using sparse representation learning to identify latent behavioral features associated with distinct activity patterns. We relate the resulting features to sleep disturbance, stress, and loneliness using generalized estimating equations with Mundlak decomposition, enabling separation of stable between-person differences from within-person changes over time. Our analysis reveals that the three outcomes are characterized by different temporal dynamics: stress is predominantly associated with persistent between-person variation, loneliness is more strongly linked to within-person fluctuations, and sleep disturbance reflects a combination of both. Importantly, these within-person behavioral signals are not recovered by conventional handcrafted network-traffic features, highlighting the advantages of learned representations for longitudinal behavioral modeling. Overall, our findings demonstrate that encrypted network traffic contains interpretable behavioral information and can support passive, scalable monitoring of behavioral dynamics, particularly changes relative to an individual's typical pattern of activity.

The Usefulness Gap in Proof-of-Useful-Work: An Empirical Study of Pearl's cuPOW Protocol

2026-06-05T21:17:13Z

Pearl, a Layer-1 blockchain with high-profile AI industry endorsements, markets its Proof-of-Useful-Work (PoUW) protocol as simultaneously securing the network and performing AI inference. We present the first systematic empirical measurement of a deployed PoUW system, finding that Pearl's 24 EH/s network -- representing approximately 320,000 GPU-equivalents consuming an estimated 112 MW -- produces zero useful AI computation. Budget GPU rental prices rose 38% and utilization surged from 57% to 94% following the mining software's public release, displacing legitimate research workloads. Our measurements span five dimensions: (1) network composition analysis of 8,012 workers shows all have inference-capable hardware, yet the dominant mining software contains no inference code; (2) the verification protocol accepts random matrices by design, confirmed by 44 pool-accepted shares from our open-source miner across NVIDIA, AMD, CPU, and Apple Silicon hardware; (3) statistical distribution checks are trivially defeated by adversarial Gaussian sampling; (4) mining economics are marginal at current PRL prices ($0.76), with ROI ranging from -1% to +67% depending on GPU tier -- near breakeven for most hardware; and (5) the mining computation is commodity integer arithmetic portable to any hardware platform, offering no vendor lock-in. These findings quantify the verifiability-usefulness tension identified theoretically by Leinweber et al., providing concrete measurements of its magnitude and economic consequences in a deployed system.

Memetic Capture: A Pluralistic Policy Framework for Governing AI-Driven Cultural Disempowerment

2026-06-05T19:32:36Z

Culture is the most insidious vector of gradual human disempowerment by AI: unlike economic or political displacement, cultural displacement attacks the very preferences and values through which humans recognise and resist disempowerment itself. We argue that existing AI governance frameworks suffer from a critical blind spot by treating cultural impact as secondary to economic and safety concerns. This paper develops \emph{memetic capture} as a unifying concept for AI-driven cultural disempowerment, and proposes the \textbf{Cultural Pluralistic Governance Framework (CPGF)}, a four-tier policy architecture combining quantitative cultural influence metrics, democratic value assemblies, pluralistic deployment standards, and transnational coordination mechanisms. We argue that pluralism is not merely an ethical requirement for such governance but a structural necessity: monocultural AI governance accelerates the very disempowerment it claims to prevent. We identify concrete policy levers, discuss implementation tensions, and outline a research agenda at the intersection of pluralistic alignment and cultural AI governance.

Bounded by Risk, Not Capability: Quantifying AI Occupational Substitution Rates via a Tech-Risk Dual-Factor Model

2026-06-05T19:13:05Z

The deployment of Large Language Models (LLMs) has ignited concerns about technological unemployment. Existing task-based evaluations predominantly measure theoretical "exposure" to AI capabilities, ignoring critical frictions of real-world commercial adoption: liability, compliance, and physical safety. We argue occupations are not eradicated instantaneously, but gradually encroached upon via atomic actions. We introduce a Tech-Risk Dual-Factor Model to re-evaluate this. By deconstructing 923 occupations into 2,087 Detailed Work Activities (DWAs), we utilize a multi-agent LLM ensemble to score both technical feasibility and business risk. Through variance-based Human-in-the-Loop (HITL) validation with an expert panel, we demonstrate a profound cognitive gap: isolated algorithmic probabilities fail to encapsulate the "institutional premium" imposed by experts bounded by professional liability. Applying a strictly algorithmic baseline via mathematical bottleneck aggregation, we calculate Relative Occupational Automation Indices ($OAI$) for the U.S. labor market. Our findings challenge the traditional Routine-Biased Technological Change (RBTC) hypothesis. Non-routine cognitive roles highly dependent on symbolic manipulation (e.g., Data Scientists) face unprecedented exposure ($OAI \approx 0.70$). Conversely, unstructured physical trades and high-stakes caretaking roles exhibit absolute resilience, quantifying a profound "Cognitive Risk Asymmetry." We hypothesize the emergent necessity of a "Compliance Premium," indicating wage resilience increasingly tied to risk-absorption capacity. We frame these findings as a cross-sectional diagnostic of systemic vulnerability, establishing a foundation for subsequent Computable General Equilibrium (CGE) econometric modeling involving dynamic wage elasticity and structural labor reallocation.

Reimagining Open Source and Openness in AI: Co-Creating Responsible Technological Futures

2026-06-05T18:23:39Z

Debates over open source and openness in artificial intelligence have intensified as policymakers, researchers, and practitioners grapple with how foundation models should be developed and governed to balance innovation, accountability, and public interest. However, there has been limited empirical work examining how diverse stakeholders collectively understand and negotiate responsible openness in AI, particularly through participatory processes that extend beyond industry-led definitions and frameworks. This paper presents findings from a multi-sectoral workshop grounded in futures thinking and participatory design methods. The workshop generated co-created visions of desirable futures and the role of AI, alongside a set of action pathways and a research roadmap focused on responsible open source and openness in AI. This paper makes three key contributions. First, it empirically documents the co-created visions, actions, and research priorities. Second, it identifies four core tensions that emerged as participants translated high-level aspirations into concrete actions, revealing conflicting interpretations of openness regarding its purpose (as an end or a means), its scope (expansion versus meaningful access), and its operation (mandatory versus conditional, sufficient versus dependent on governance and use). These tensions illustrate that responsible openness is not a singular technical solution, but a negotiated sociotechnical project shaped by values, positionalities, and priorities. Third, the paper advances methodological approaches in AI governance by demonstrating how participatory futures methods can surface plural visions, actions, and research priorities that extend beyond dominant, largely corporate, narratives, offering empirical insight into how openness, power, and accountability are negotiated in practice.