arXiv subjects

Xitong Li

Publications and source records attributed to Xitong Li.

5 recordsLinked to original sources

arXiv2026

The Double-edged Effect of Banning Generative AI on Online Question-and-Answer Communities: Evidence from Stack Exchange

We investigate how banning generative artificial intelligence-generated content (AIGC) affects knowledge seeking, knowledge contribution, and contribution efficiency in online question-and-answer communities. After the launch of ChatGPT in late November 2022, several Stack Exchange communities implemented official bans on AIGC over concerns such as less reliable and socially engaged content. Leveraging data from the full network of Stack Exchange communities, we employ a difference-in-differences (DID) approach to examine the impacts of these bans. Our results reveal a double-edged impact: while the AIGC ban increases knowledge seeking, as evidenced by a higher volume of posted questions, it simultaneously reduces contribution efficiency, reflected in a lower proportion of questions receiving satisfactory answers within the expected time frame. Notably, these impacts are only evident in non-STEM communities. We take a socio-technical perspective to explore information reliability and social interactivity as two plausible underlying factors driving the observed changes. Our mechanism exploration reveals that the AIGC ban spurs question volume in topics where AIGC is less reliable and where social interaction is highly expected. In contrast, the ban hampers answer efficiency in communities where LLMs are capable of producing reliable answers and where social interactivity is minimal. Additionally, our results indicate the increased human involvement from knowledge seekers and contributors following the ban. They adapt their behavior by posting questions and answers that are more informationally rich and socially engaging. Overall, our findings offer actionable implications for platform managers, community moderators, and policymakers of online Q&A communities.

cs.CY↗

arXiv2026

Large Language Models Polarize Ideologically but Moderate Affectively in Online Political Discourse

The emergence of large language models (LLMs) is reshaping how people engage in political discourse online. We examine how the release of ChatGPT altered ideological and emotional patterns in Reddit's largest political forum. Analysis of millions of comments shows that ChatGPT intensified ideological polarization: liberal-leaning authors posted increasingly liberal comments, while conservative-leaning authors posted increasingly conservative comments. Multiple falsification tests suggest that these findings are unlikely to be driven by contemporaneous events, such as the 2022 U.S. midterm elections, or by broader platform-wide trends in political polarization. Mechanism tests show that this shift does not stem from the creation of more persuasive or ideologically extreme original content using LLM. Instead, it originates from the tendency of LLM-assisted comments to echo and reinforce the original post's viewpoint, a pattern consistent with algorithmic sycophancy. Yet, despite growing ideological divides, affective polarization, measured by hostility and toxicity, declined. These findings reveal that LLMs can simultaneously deepen ideological separation and foster more civil exchanges, challenging the long-standing assumption in literature that extremity and incivility necessarily move together.

econ.GN↗

arXiv2025

Few-Shot Query Intent Detection via Relation-Aware Prompt Learning

Intent detection is a crucial component of modern conversational systems, since accurately identifying user intent at the beginning of a conversation is essential for generating effective responses. Recent efforts have focused on studying this problem under a challenging few-shot scenario. These approaches primarily leverage large-scale unlabeled dialogue text corpora to pretrain language models through various pretext tasks, followed by fine-tuning for intent detection with very limited annotations. Despite the improvements achieved, existing methods have predominantly focused on textual data, neglecting to effectively capture the crucial structural information inherent in conversational systems, such as the query-query relation and query-answer relation. To address this gap, we propose SAID, a novel framework that integrates both textual and relational structure information in a unified manner for model pretraining for the first time. Building on this framework, we further propose a novel mechanism, the query-adaptive attention network (QueryAdapt), which operates at the relation token level by generating intent-specific relation tokens from well-learned query-query and query-answer relations explicitly, enabling more fine-grained knowledge transfer. Extensive experimental results on two real-world datasets demonstrate that SAID significantly outperforms state-of-the-art methods.

cs.CL↗

arXiv2025

My Advisor, Her AI and Me: Evidence from a Field Experiment on Human-AI Collaboration and Investment Decisions

Amid ongoing policy and managerial debates on keeping humans in the loop of AI decision-making, we investigate whether human involvement in AI-based service production benefits downstream consumers. Partnering with a large savings bank in Europe, we produced pure AI and human-AI collaborative investment advice, passed it to customers, and examined their advice-taking in a field experiment. On the production side, contrary to concerns that humans might inefficiently override AI output, we find that giving a human banker the final say over AI-generated financial advice does not compromise its quality. More importantly, on the consumption side, customers are more likely to follow investment advice from the human-AI collaboration compared to pure AI, especially when facing riskier decisions. In our setting, this increased reliance leads to higher material welfare for consumers. Additional analyses from the field experiment and an online experiment show that the persuasive power of human-AI advice cannot be explained by consumers' beliefs about enhanced advice quality due to human-AI complementarities. Instead, the benefit stems from human involvement acting as a peripheral cue that increases the advice's affective appeal. Our findings suggest that regulations and guidelines should adopt a consumer-centered approach by fostering service environments in which humans and AI systems can collaborate to improve consumer outcomes. These insights are relevant for managers designing AI-based services and for policymakers advocating for human oversight in AI systems.

econ.GN↗

arXiv2018

The Definition and Numerical Method of Final Value Problem and Arbitrary Value Problem

Many Engineering Problems could be mathematically described by Final Value Problem, which is the inverse problem of Initial Value Problem. Accordingly, the paper studies the final value problem in the field of ODE problems and analyses the differences and relations between initial and final value problems. The more general new concept of the endpoints-value problem which could describe both initial and final problems is proposed. Further, we extend the concept into inner-interval value problem and arbitrary value problem and point out that both endpoints-value problem and inner-interval value problem are special forms of arbitrary value problem. Particularly, the existence and uniqueness of the solutions of final value problem and inner-interval value problem of first order ordinary differential equation are proved for discrete problems. The numerical calculation formulas of the problems are derived, and for each algorithm, we propose the convergence and stability conditions of them. Furthermore, multivariate and high-order final value problems are further studied, and the condition of fixed delay is also discussed in this paper. At last, the effectiveness of the considered methods is validated by numerical experiment.

math.NA↗