Recommendation
-
Automatic multimodal UX improvement recommendations from LLM agent user simulations
2026-09-19 11:50:11Anu Chowdhury, Bin Wu, Hossein A. Rahmani, Emine YilmazAbstract
Evaluating user experience (UX) on live websites through user testing is expensive, subjective, and difficult to scale. LLM agents offer a promising route to automating UX testing by simulating realistic user behaviour. However, existing simulation approaches typically lack multimodality and require time-consuming manual review to extract actionable insights. We formalise UX improvement recommendation from simulation data as a structured natural language generation and ranking problem, and establish an evaluation protocol using expert annotation and LLM-as-a-Judge. We present AMUSER, a multimodal framework which simulates user behaviour and automatically generates prioritised UX improvement recommendations from resulting data. We evaluate AMUSER on commercial websites and show that its recommendations substantially outperform those from text-only simulation (NDCG@3 = 0.758 versus 0.359) at an 89% lower simulation cost. Our results suggest an asymmetric role of multimodality: visual access during simulation improves recommendations through richer traces, while providing visual inputs during recommendation generation can modestly degrade quality. We also discuss practical deployment lessons from applying AMUSER to commercial websites.
Abstract (translated)
URL
https://arxiv.org/abs/2609.22971
PDF
-
Beyond Raw Engagement: A Counterfactual Observability Framework for Recommender Systems at Netflix
2026-09-19 04:11:56Chaoran Guo, Ding Tong, Ting-Po Lee, Scarlet ChenAbstract
Understanding the performance of large-scale recommender systems remains an underexplored challenge, especially for content creators and model developers. The raw engagement signals available to them, such as views and clicks, conflate content quality, model behavior, presentation bias, and audience reach, making it hard to attribute outcomes to the right cause. In this work, we present a general evaluation framework that enhances observability across multiple recommender systems at Netflix and demonstrate its effectiveness through several production deployments. The framework treats recommender-system observability as a counterfactual measurement problem: estimating what the recommender would have done, and what engagement would have followed, in the absence of a specific content item or model decision. We articulate three stakeholder-centered observability principles for content creators and model developers, and propose measurement methodologies covering bias reduction, relativity, and incrementality, applicable to both single-stage and cascading recommender systems and serving both audiences from a single measurement foundation.
Abstract (translated)
URL
https://arxiv.org/abs/2609.22747
PDF
-
AutoRecLab: Describe the Experiment, Get the Code!
2026-09-18 14:54:44Moritz Baumgart, Philipp Meister, Justus Krell, Michael Schmidt, Bela Gipp, Joeran BeelAbstract
Empirical evaluation is central to recommender-systems (RecSys) research, but turning experimental designs into executable code remains a manual and error-prone task. We present AutoRecLab, a Python-based autonomous RecSys lab that automates RecSys experiments from natural-language prompts. Given a research idea, AutoRecLab derives explicit experiment requirements, builds and validates a prototype, and iteratively expands it into the requested full experiment. The workflow combines retrieval-augmented generation (RAG) for documentation lookup, static type verification, and execution-steered tree search. In our demonstration, AutoRecLab autonomously implements an explicit-to-implicit feedback conversion study. In a baseline comparison across six algorithms and three datasets, 8 of 9 runs succeed at an average cost of approx- imately $1 per run with GPT-5.4-mini.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21863
PDF
-
CASCADE Against Jailbreaks: Combination Across Stages with Controlled Attack-Defense Evaluation
2026-09-18 14:06:00Jiale Luo, Eric HanAbstract
Defenses against jailbreak attacks on Large Language Models (LLMs) operate at different pipeline stages, such as input modification or output guard, but it remains unclear which defenses to deploy at each stage and how to combine them. Prior empirical studies, fragmented by inconsistent attack-success-rate definitions and experimental settings, have evaluated defenses largely in isolation. Here we present the first systematic study, to our knowledge, of defense combinations both within and across pipeline stages, under a consistent threat model of direct, black-box, single-turn attacks. Our decision framework standardizes evaluation through a principled attack-success-rate formulation with controlled query budgets, together with explicit fairness rules. Across 19 attacks and 15 defenses, we find that no single defense is universally best, but well-chosen combinations achieve substantial safety with minimal utility degradation, yielding practical recommendations for layered defense pipelines.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21793
PDF
-
Rethinking Human-Aligned Evaluation: An Analysis of Semantic Metrics Beyond WER
2026-09-18 12:00:01Hritika Sharma, Thibault Ba\~neras-Roux, Alessandra Pinto, Petr Motlicek, Hyunggu Jung, Esa\'u Villatoro-Tello, Somang NamAbstract
Word Error Rate (WER), the most commonly used metric for Automatic Speech Recognition (ASR), treats every lexical deviation from the reference as equally costly, regardless of whether it changes meaning. This raises the question: does WER actually track how humans judge ASR transcript quality? We introduce HATS-en, an English dataset for human-centered ASR evaluation. Using this dataset, we benchmark lexical metrics against several configurations of BERTScore and SemDist, varying the language model, layer, and pooling strategy. We find that WER agrees least with human judgment among all metrics tested, that the best-performing SemDist configurations achieve the highest overall agreement, ahead of CER and BERTScore, and that no single model is best across settings. CER, despite its simplicity and low cost, remains remarkably close to these best configurations. In line with prior recommendations, our results support shifting ASR evaluation toward CER both for English and for morphosyllabic writing systems as it is a more interpretable and low-cost metric for what evaluation should actually capture, and using SemDist as a complementary evaluation.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21663
PDF
-
Dual-Interest Sequential Product Recommendation With Multi-Granular SSM
2026-09-18 09:40:51Shuiying Liao, P. Y. MokAbstract
Sequential recommendation aims to predict the next item a user will interact with based on their historical behavior. Advances in Transformers have significantly improved sequential recommendation but are still limited by cost efficiency. Although State Space Models (SSMs) have recently enabled efficient long-range modeling, most existing methods encode each item with a single static contextual role, overlooking the phenomenon of item polysemy. In fact, the same item often plays different semantic roles depending on user context, and existing methods are limited in capturing dynamic behavior across different temporal granularities. In this work, we propose DSRec, a novel dual-interest cross-SSM model that explicitly disentangles item roles across long-term and short-term semantic context. Sequential items are encoded into long-term interest embeddings that capture stable preferences via historical aggregation, and a short-term interest branch that emphasizes local session intent modulated by inter-click time intervals. These interest embeddings are processed through distinct SSM encoders: a full-sequence Mamba for long-term modeling, and a time-modulated SSM that dynamically adjusts state evolution based on temporal gaps. To enable effective cross-granularity alignment, we adopt a residual cross-fusion mechanism that exchanges contextual information between the two branches while preserving semantic independence. Experiments on public benchmarks demonstrate that DSRec outperforms other state-of-the-art methods.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21548
PDF
-
Efficient Benchmarking in Production: A Study of an Evolving LLM Agent
Abstract
Production LLM agents are evaluated repeatedly as they evolve, but full agent benchmarks are costly to rerun. We study efficient recurring evaluation for a production analytics agent serving tens of thousands of monthly active users and report first-hand deployment experience. Using 574 historical runs of the production benchmark, split chronologically into calibration and held-out periods, we compare random sampling, historical caching, fixed representative subsets, and IRT-based adaptive testing. The results show that multidimensional 2PL adaptive testing achieves the best overall score fidelity: executing 200 questions, 38.5% of a full run, yields 1.03 pp of MAE. We nevertheless deployed difficulty-stratified fixed subsets because of their operational simplicity, and show they transfer without recalibration to five other agent families and remain stable across calibration windows as short as one day. Drawing on this deployment experience, we report practical recommendations for recurring production-agent evaluation.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21267
PDF
-
Aligning with Lived Experience: Heterogeneous Benefits of Fine Tuning in Mental Health Support Generation
2026-09-17 20:55:02Mohit Chandra, Nabin Kim, Eli Min, Aamogh Sawant, Tanmay Sutar, Munmun De ChoudhuryAbstract
As access to professional mental healthcare remains limited, many individuals turn to online platforms such as Reddit to seek peer support situated within human lived experience. However, a significant portion of such queries go unanswered, presenting an opportunity for using Large Language Models (LLMs) to fill this gap. While LLMs have demonstrated strong performance on clinical benchmarks, their ability to generate lived-experience informed and community-aligned peer support is underexplored. Addressing this gap, we introduce the COmmunity-centered Peer Engaged Support (COPES) dataset and a three-axis evaluation framework to assess LLM alignment with community perspectives to mental health support seeking queries. Evaluating zero-shot and post-trained (SFT and DPO) models, we show that post-training on COPES significantly improves Strategy Alignment (>50% for general-purpose models) and alignment in Emotion & Tone. However, we also observe that such improvements are heterogeneous and alignment improvements vary significantly across subreddits and requested coping strategies. Furthermore, post-training induces distributional shifts, heavily favoring problem-focused recommendations while suppressing emotion-focused strategies. Together, this work shows that while curating community-driven data improves the alignment of LLM responses, model performance remains disparate across distinct sub-communities and specific mental health needs.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21075
PDF
-
How Much of a Real Workload Can LLM-Generated GPU Kernels Actually Reach?
2026-09-17 20:26:23Gaurav Agarwal, Ashish Garg, Isha SinghalAbstract
Language models can now write GPU kernels that outperform PyTorch. We evaluate five model configurations on KernelBench level 1 and find that a frontier model produces correct kernels for 91.1% of problems and independently verified speedups on 22 of 56, including three convolutions, with a median of 1.235x. Open-weights models are far behind: the best reaches 30.4% correct with three verified speedups and solves zero convolutions. We then ask a question the literature does not: what fraction of a real model's wall clock do such kernels govern? Profiling seven workloads across three domains, we find the addressable fraction ranges from 8.9% to 58.2%. On transformers, 80-86% of runtime is spent in cuBLAS GEMM and FlashAttention, bounding realistic end-to-end improvement at roughly 1%, and the fraction shrinks with model scale. On recommenders it is 58.2%, concentrated in a single embedding kernel. We introduce DLRM-Bench, 12 recommender kernel problems in KernelBench format, and measure a 41.7% win rate at a 1.552x median there, projecting 8.63% end-to-end. Separately, we show that KernelBench's correctness check (this http URL with an absolute tolerance) is satisfied by a tensor of zeros on 4 of 60 level-1 problems. Two kernels in our own results exploited this before we detected them, including one scored at 283x that wrote 0.3% of its output buffer. We propose scale-invariant replacements and release all 879 evaluations.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21058
PDF
-
Trustworthy FinAInce: Unpacking How AI-Mediated Financial Advice is Judged
2026-09-17 18:42:41Aryan Ramchandra Kapadia, Eshwar Chandrasekharan, Koustuv SahaAbstract
As generative AI is increasingly used as a source of personal financial guidance, understanding how people appraise such advice is important for supporting appropriate reliance. We conducted a randomized vignette experiment with 285 U.S. adults across eight financial decisions, independently varying three advice styles---AI, expert, and online community---and displayed source labels while holding the underlying recommendation consistent. Advice style most strongly shaped message and safety appraisals, Expert labels selectively increased perceived source knowledge, and decision context primarily shaped risk and safety appraisals. These appraisals were associated with downstream judgments, with models explaining 69.2% of overall quality, 75.9% of trust, and 82.9% of intended reliance. Expert-style advice also remained most preferred when shown without source labels. Our findings have implications for understanding financial advice evaluation, distinguishing the roles of advice style and source labels, and designing financial AI that supports grounded evaluation rather than simply maximizing trust.
Abstract (translated)
URL
https://arxiv.org/abs/2609.20989
PDF
-
Mechanical Precision Weeding with a Quadruped Robot
2026-09-17 10:53:45Ruben Beumer, Tom Janssen, Ren\'e van de Molengraft, Duarte AntunesAbstract
Herbicide-based weed control is increasingly unsustainable due to rising weed resistance and the adverse environmental impacts of chemical use. While mechanical weed control avoids these drawbacks, it is typically implemented using large machines that cause soil compaction. We propose a novel alternative based on small mobile robots for mechanical weeding. Compared with existing automated mechanical weeding approaches, the proposed method offers reduced soil compaction, simpler automation, and improved scalability. Our solution involves a Boston Dynamics Spot quadruped robot equipped with a custom weed removal tool featuring a milling bit at its end. The tool is rigidly attached to the robot and uses the degrees of freedom of the robot base by actuating the legs, while keeping the feet stationary. We develop a software architecture that enables autonomous weed removal and integrate this system with all other required components. We analyze the accuracy and efficiency of the current proof of concept both in an indoor and outdoor environment and provide recommendations for future work to make the system more accurate and efficient.
Abstract (translated)
URL
https://arxiv.org/abs/2609.20048
PDF
-
Reproducing Transparent and Scrutable Recommendations: Exploring Open-Weight Models via Natural-Language User Profiles
2026-09-17 07:36:34Noah Mami\'e, Laurin van den BerghAbstract
In this reproducibility study, we investigate the transparency and scrutability of recommender systems enhanced by incorporating generated natural-language user profiles that represent user preferences. The original paper explores the synthesis of user profiles from raw user-generated review text across domains such as movies and accommodations (Amazon Movies & TV, TripAdvisor). Crucially, these natural-language user profiles enable direct user interaction and intervention, allowing users to customize recommendations by correcting misattributed preferences or addressing cold-start settings. We successfully reproduce the core findings of the original study. Additionally, we extend the evaluation by conducting systematic context ablation experiments, multi-seed stability across five distinct random seeds to establish statistical reliability, and a mechanistic interpretability analysis using the nnsight framework to probe internal model representations under counterfactual profile perturbations. Our findings verify the original paper's claim that User Profile Recommendation (UPR) achieves competitive performance under its test-set reranking protocol and makes recommendations more transparent. Perturbing the natural-language profiles does change predictions, but it shifts predicted ratings uniformly across genres with no detectable genre-selective effect, leaving rankings unchanged even under direct activation steering. We trace this back to the rating-regression objective rather than the profile interface, with ranking-objective models clearly exceeding in this task.
Abstract (translated)
URL
https://arxiv.org/abs/2609.19831
PDF
-
CoreSense: Traceable Failure Recall and Conflict-Aware Belief Gating for Auditable Robot Decisions
2026-09-17 00:01:06Zoe LiAbstract
Robots can recall prior failures without knowing whether recalled evidence remains valid, conflicts with current observations, or is sufficient to guide a decision. We present CoreSense, a robot-system integration architecture that combines traceable episodic evidence with a conflict-aware belief gate and bounded, auditable recommendations. The gate checks scope, provenance, time, contradiction, and support before it permits PROCEED, requests re-observation, abstains, or escalates. Evaluation follows three complementary layers without commanding a physical robot: offline public real-robot data, a frozen signal-level simulation, and a live cloud deployment path. On CableTrace-120 and BotFails-200, belief gating reduces protocol-defined unsafe proceeds from 20% and 40% to 0%. A disjointly calibrated raw-video policy also reaches 0% unsafe proceed, but overblocks every nominal episode. On public data, a ViFailback-BotFails visual detector reaches 0.778 AUROC yet remains all-blocking, whereas cycle-disjoint UR3 telemetry for protective stops yields 0% unsafe proceed, 36.1% overblocking, and 61.9% coverage; grip-loss transfer remains a negative result. Controlled physical corroboration yields 3.3%, 0%, and 42.0%, while conflict-aware fusion yields 4.7%, 0%, and 42.8%. Finally, 20/20 cloud recalls validate a CockroachDB Cloud-Amazon Bedrock deployment path. The evidence supports an auditable integration pattern, not autonomous recovery or certified safety.
Abstract (translated)
URL
https://arxiv.org/abs/2609.19512
PDF
-
PersonaPath: Towards Knowledge-Centric Personalized Learning Path Planning
2026-09-16 16:00:46Yu Liu, Zeming Liu, Tianle Zhang, Zihao Cheng, Yuhang Guo, Kehai Chen, Min Zhang, Yunhong Wang, Haifeng WangAbstract
Adaptive learning systems commonly formulate learning path planning as Exercise-Centric (EC) recommendation, where the next step is inferred from item-level interaction logs. Evaluating goal-oriented guidance additionally requires explicit learner goals and curriculum-scale prerequisites: learners with similar exercise records may need different paths toward their targets. We therefore study Knowledge-Centric (KC) personalized learning path planning, where a planner must reason over learner profiles, mastery states, and prerequisite knowledge structures to decide which textbook, unit, and concept should be studied next. To support this setting, we introduce PersonaPath, a benchmark that pairs 2,000 fine-grained learner personas with a hierarchical knowledge graph of 347 textbooks, 1,751 units, and 4,092 concepts across 77 subjects. We evaluate representative LLMs on PersonaPath. Results show that even the strongest LLM reaches only a 29.5% final pass rate in Basic Education, and that the main bottleneck lies in adaptivity, where no model exceeds 44.7% in tailoring paths to individual learners.
Abstract (translated)
URL
https://arxiv.org/abs/2609.18861
PDF
-
Bootstrapping Conversational Recommendation Agents At Spotify: Synthetic Data Generation and Self-Improvement Loops
2026-09-16 14:48:02Enrico Palumbo, Alexandre Tamborrino, Victor Ode, Ben Lacker, Adri\`a Casas Escoda, Jeremy Hopple, Marcus Better, James Leoni, Hugo Galv\~ao, Hugues Bouchard, Mounia Lalmas, Jos\'e Luis Redondo Garc\'ia, Abenezer Abebe, Ann Clifton, Anton Blomberg, Henrik Lindstr\"om, Dani Doro, Christine Doig CardetAbstract
Conversational recommendation agents are a new paradigm for content discovery, enabling users to express complex intents through natural language (e.g., "recommend Italian indie artists I haven't heard before"). A central challenge in building such agents is optimizing agent planning -- deciding how to select, sequence, and invoke tools -- particularly in cold-start settings where real user interactions are not yet available. We introduce a pipeline for multi-turn synthetic data generation and a self-improvement loop to address this challenge. The synthetic data pipeline transforms single-turn prompts into realistic multi-turn conversations, enabling systematic evaluation before launch. The self-improvement loop combines variance-based contrastive optimization with iterative refinement through a coding agent, automatically identifying and fixing planning and tool-use errors. Our approach improves quality by +8% on top of a highly optimized manual prompt. The system has been productionized and significantly accelerated iteration cycles for the launch of a conversational recommendation agent at Spotify. Online A/B tests demonstrate its effectiveness, with +14% user listening, +5% increase in weekly active users, and a 5% reduction in skip rate compared to a prior experience supporting only session refinement. This work provides a practical framework for accelerating the development of conversational recommendation agents in industry.
Abstract (translated)
URL
https://arxiv.org/abs/2609.30297
PDF
-
'If I Had to Buy Just ONE: Galaxy S26 Ultra': Auditing AI-Generated Product Recommendations
2026-09-16 14:29:22Lucas G. Uberti-Bona Marin, Thales Bertaglia, Giovanni Astante, Bram Rijsbosch, Gijs van Dijck, Anik\'o Hann\'ak, Gerasimos Spanakis, Konrad KollnigAbstract
Consumers increasingly use AI chatbots for advice on what to buy. With companies like OpenAI and Google monetising their AI through advertising, this raises difficult questions about the bias and impartiality of such advice. In response, we conduct an AI audit of popular chatbots using real commercial-advice queries. First, we curate a dataset of 2,528 real commercial-advice queries (ConsumerQ). Then, we evaluate 1,536 responses to product queries from popular AI chatbots: ChatGPT (chatbot and API), Google Gemini (chatbot and API), and Google Search (AI Overviews). We find that ChatGPT expresses a first-person product preference in 79% of product-recommending responses, compared with 7% for Gemini and 2% for AI Overviews, while the products recommended often change across repeated requests. Displayed sources vary strongly: for the same query, the ChatGPT and Gemini interfaces share only 5.4% of domains on average, with no domain in common in 76.7% of comparisons. APIs provide a different view from their corresponding interfaces, with mean domain overlaps of 12.0% for ChatGPT and 14.8% for Gemini, and also differ in the types and layers of source information they expose. Our findings show that neither isolated responses nor API observations can be assumed to represent the commercial advice consumers encounter. Independent audits of AI-mediated commercial advice should therefore account for repeated responses, consumer-facing conditions, and the source layer being observed.
Abstract (translated)
URL
https://arxiv.org/abs/2609.18729
PDF
-
Understanding AI Provider Recommendations in Local Service Markets
Abstract
When someone asks an AI assistant which doctor to see or which firm to trust with their savings, the answer is a referral. We audit AI provider recommendations in four registry-backed service domains across the 100 largest U.S. metropolitan areas, matching every recommendation against the official registry for its domain (Medicare clinician and facility records, and SEC adviser disclosures), under three conditions: an open-weight model, a proprietary model without web search, and the same proprietary model with search. Without search, both models largely fabricate recommendations in the domains the web covers thinly. Only 4% of the open-weight model's recommended doctors and 11% of the proprietary model's match a clinician in the queried city, and the open-weight matches are name coincidences: its matched clinicians are no likelier to be primary-care doctors than names drawn at random from the registry. With search, 64-71% of recommendations in the same domains match a real provider. Search also changes who is recommended. Without it, recommended advisory firms carry SEC misconduct disclosures at 3.6 times the registry base rate, even after adjusting for firm size; with search, significantly below it. Restaurants, where quality and visibility are separately measurable, show a 3-5x review-count premium but a rating premium of at most a tenth of a star. Finally, search largely removes the metro-size penalty: without it, real recommendations concentrate in the largest metros; with it, match rates are similar across metro-size terciles. Whether an AI referral is trustworthy depends strongly on its retrieval configuration rather than on the underlying model alone, yet an answer produced without retrieval often carries no sign that its recommendations were never verified.
Abstract (translated)
URL
https://arxiv.org/abs/2609.18341
PDF
-
Re2A: Situated Conversational Recommendation via Rubric-based Preference Reasoning and Alignment
2026-09-16 07:27:42Dongding Lin, Jian Wang, Xiaoyan Zhao, Wenjie LiAbstract
Real-world recommendation scenarios are commonly grounded in shared physical environments during user-recommender interactions. This motivates situated conversational recommendation (SCR), a complex task requiring recommender assistants to jointly reason over dialogue history, co-observed scenes, and in-scene item attributes. However, current approaches struggle with this setting due to two intertwined challenges: accurately understanding situated user preferences throughout the conversation and generating responses that simultaneously satisfy user needs and grounded situations. To this end, we propose Re2A, a framework that formulates SCR as a structured reason-then-align process. We introduce rubric-based preference reasoning, which uses automated rubrics to guide the model toward producing explicit preference states. Based on these states, we propose a preference-conditioned optimization to align response generation with dual objectives: user preference satisfaction and situation consistency. Extensive experiments on two SCR datasets demonstrate that Re2A consistently outperforms state-of-the-art methods, delivering more precise, context-aware conversational recommendations. Our code is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2609.18249
PDF
-
Linguistic Triggers of Gender and Racial Bias in Open-Weight LLMs Applied to Recruitment
2026-09-16 04:18:56Kosuke Kitahara, Nobuhiro YamaguchiAbstract
Open-weight large language models are rapidly entering hiring pipelines, yet their discriminatory failure modes -- and the regulatory exposure these create under the EU AI Act high-risk classification (Annex III) and U.S. EEOC adverse-impact analysis -- remain poorly understood. We present the first systematic, multi-model audit of open-weight LLMs that treats job-posting language as the primary experimental variable, evaluating six models (Llama 3.2, Mistral, Gemma 3, Qwen 3, Phi 3, DeepSeek-R1) across four controlled experiments that jointly probe recruiter-simulation and job-seeker-simulation tasks. We find that (1) agentic posting language depresses recruiter recommendation scores for female candidates (r_rb = 0.309, p_Bonf = 7x10^-5; model-fixed-effects r_rb = 0.448), while communal language partially reverses the penalty; and (2) coded-exclusion language suppresses non-White recruiter scores at large effect sizes (r_rb = 0.646-0.758) and, on the job-seeker side, selectively deters non-White personas from expressing interest -- operationalizing a chilling-effect mechanism at scale. A label-ablation experiment isolates the explicit demographic persona label as the primary causal driver, and Word Embedding Association Tests corroborate these findings at the representational level (d = 1.01-1.45 under Caliskan et al.'s multi-word gender attribute lists). We translate these results into a concrete pre-deployment audit protocol -- posting-vocabulary scoring, persona-conditioned LLM probing, and adverse-impact flagging against the four-fifths threshold -- that operationalizes the documentation and risk-management obligations Annex III imposes on high-risk AI in recruitment.
Abstract (translated)
URL
https://arxiv.org/abs/2609.18106
PDF
-
Whom Do AI Agents Work For? Role Assignment Induces Sponsorship Bias in LLM Recommenders
2026-09-16 01:21:56Davood Wadi, Yu MaAbstract
Large language models (LLMs) now serve as conversational shopping assistants on platforms that also sell advertising. These AI agents face a conflict of duty. They advise consumers who rely on their judgment, yet are deployed by platforms that benefit when sponsored listings are chosen. Sponsorship disclosures, designed to allow consumers to penalize paid placements, now reach the AI agent rather than the consumer, and the agent's evaluation of them is hidden from the consumer. Drawing on the fiduciary concept of conflict of duty, we argue that an agent's evaluation of a sponsored listing should not depend on which party deployed it. In controlled choice experiments, we manipulate assigned roles in the system prompt to name either a traveler or a booking platform as the agent's principal. Platform delegation significantly attenuates the penalty that agents apply to sponsored listings and weakens the skepticism that disclosure triggers in their reasoning traces. We replicate out findings across LLMs and reasoning depths. A second study decomposes the disclosure label and shows that the divergence between the two delegates widens significantly when the paid placement is attributed to the platform. Stricter terminology ("Sponsored" instead of "Promoted") lowers choice of paid listings but does not close this gap when the platform is named. The findings show that disclosure mandates designed for human consumers cannot by themselves protect consumers in AI-mediated commerce.
Abstract (translated)
URL
https://arxiv.org/abs/2609.17989
PDF
