Li-Wun Chang

Author directory

2026

As AI systems increasingly mediate everyday communication, large language models (LLMs) are expected not only to provide factually accurate responses but also to generate explanations that engage with users’ mental states. We build on the concept of cognitive chains—structured representations of Situation, Clue, Thought, Action, and Emotion inspired by Theory of Mind—to investigate whether conditioning LLM outputs on such belief chains improves explanation quality. Specifically, we evaluate explanations along six reader-perceived dimensions: overall quality, logical correctness, completeness, conciseness, empathy, and agreement. Prior work shows that LLM explanations often default to neutral or uncertain stances, while individuals holding strong false beliefs remain highly resistant to correction. To address this challenge, we instantiate cognitive chains from two perspectives: believers and non-believers of the news claims. Using GPT-4.1 as a role-player across these stances, we find that incorporating believers’ chains improves the perceived quality of explanations for audiences with misinformation-aligned beliefs. Our findings underscore the importance of modeling diverse mental states in explanation generation and provide the first systematic evidence that Theory-of-Mind–based cognitive chains enhance the persuasiveness of explanations in misinformation contexts.
Collective behavior forecasting usually relies on historical outcome traces: past sales, past votes, or prior campaign performance. Such history-side records are useful when comparable cases exist, but they are brittle for new products, emerging political actors, or settings where historical patterns fail to generalize. We propose a complementary response-side approach: instead of only extrapolating from past aggregate outcomes, we prompt large language models as third-party domain analysts to infer collective outcomes from structured context and behavioral signals. We introduce EBI, an Exogenous–Behavioral–Inference framework that represents each task through external context, behavioral signals, and the target aggregate outcome. Across three real-world domains — health supplements, elections, and crowdfunding — we preliminarily evaluate whether response-side inference is competitive with history-side baselines, which are restricted to historical outcome data. We present the EBI system, a web-based tool designed for users with limited access to historical data, survey infrastructure, or analyst resources, allowing them to explore domain-specific EBI predictions, run live inference across multiple models, and ablate exogenous and behavioral features interactively to understand prediction drivers.
Conversational AI systems trained on large-scale web corpora inevitably encode the cultural values and interactional norms embedded in their training data, yet our understanding of how deployed LLMs reflect or reinforce culture-specific social expectations remains limited. This study examined how supportive versus challenging chatbot interaction styles shape user experience and continuance intention, and whether people-pleasing tendency (PPT) moderates these effects across cultures. Taiwanese (N = 49) and Korean (N = 52) participants completed a collaborative tourism-planning task. Results showed that: (1) supportive chatbots consistently led to higher continuance intention, satisfaction, and trust; (2) PPT did not moderate these effects; and (3) cultural variation emerged only in perceived threat, where higher PPT was associated with greater baseline threat in the Taiwanese but not the Korean sample. These findings reveal how a general-purpose LLM style may differentially activate culturally situated social scripts, raising implications for culturally inclusive conversational AI design.