Tag Archives: generative AI

Large Language Models in the Institutional Press: Investigating the Effects on Information Sourcing and News Production

Zhang, Xiaoke, Myunghwan Lee, Mi Zhou, Gene Moo Lee. “Large Language Models in the Institutional Press: Investigating the Effects on Information Sourcing and News Production,” 3rd round R&R, MIS Quarterly.

  • Presentations: UBC (2024), DS (2024), CIST (2024), BIGS (2024), JUSWIS (2025), UIUC (2025)
  • Industry partner: Muhayu

Large language models (LLMs) are transforming journalism by directly entering journalistic workflows, introducing new opportunities and challenges for the institutional press. This study investigates how LLM assistance affects journalists’ information sourcing in news production using a mixed-method approach. We begin with a qualitative study of 43 journalists to identify and theorize how LLM assistance affects three core journalistic values: publication promptness, information source quantity, and information source originality. We then compile a large-scale dataset of 1,073,742 news articles from 111 South Korean news outlets and collaborate with industry experts to detect undisclosed LLM-assisted articles. Our event-level analysis shows that LLM assistance accelerates publication but reduces the number of information sources used in news articles, with a larger decline in primary sources than in secondary sources. Heterogeneity analyses and a randomized experiment suggest that this reduction is driven by two mechanisms: an LLM generation mechanism that narrows the set of retrieved and represented sources, and a metacognitive regulation mechanism that reduces journalists’ active search and evaluation. We further show that these effects extend beyond individual articles. A journalist-level difference-in-differences analysis indicates that LLM adoption leads to persistent reductions in source usage over time. Our findings offer practical implications for LLM system design, newsroom practices, and institutional disclosure policy.

Xiaoke Zhang’s Master’s Thesis

Xiaoke Zhang (2023). “How Does AI-Generated Voice Affect Online Video Creation? Evidence from TikTok”, Master’s Thesis, University of British Columbia

Supervisors: Gene Moo Lee, Mi Zhou

The rising demand for online video content has fostered one of the fastest-growing markets as evidenced by the popularity of platforms like TikTok. Because video content is often difficult to create, platforms have attempted to leverage recent advancements in artificial intelligence (AI) to help creators with their video creation process. However, surprisingly little is known about the effects of AI on content creators’ productivity and creative patterns in this emerging market. Our paper investigates the adoption impact of AI-generated voice – a generative AI technology creating acoustic artifacts – on video creators by empirically analyzing a unique dataset of 4,021 creators and their 428,918 videos on TikTok. Utilizing multiple audio and video analytics algorithms, we detect the adoption of AI voice from the massive video data and generate rich measurements for each video to quantify its characteristics. We then estimate the effects of AI voice using a difference-in-differences model coupled with look-ahead propensity score matching. Our results suggest that the adoption of AI voice increases creators’ video production and that it induces creators to produce shorter videos with more negative words. Interestingly, creators produce more novel videos with less self-disclosure when using AI voice. We also find that AI-voice videos received less viewer engagement unintendedly. Our paper provides the first empirical evidence of how generative AI reshapes video content creation on online platforms, which provides important implications for creators, platforms, and policymakers in the digital economy.

 

AI Voice in Online Video Platforms: A Multimodal Perspective on Content Creation and Consumption

Zhang, Xiaoke, Mi Zhou, Gene Moo Lee “AI Voice in Online Video Platforms: A Multimodal Perspective on Content Creation and Consumption,” 3rd round R&R at MIS Quarterly.

  • Best Student Paper Nomination at CIST 2023; Best Paper Runner-Up Award at KrAIS Summer Workshop 2023
  • Presentations: INFORMS DS (2022), UBC (2022), WITS (2022), Yonsei (2023), POSTECH (2023), ISMS MKSC (2023), CSWIM (2023), KrAIS Summer (2023), Dalhousie (2023), CIST (2023), Temple (2024), Santa Clara U (2024), Wisconsin Milwaukee (2024)
  • Media coverage: [UBC News] [Global News]
  • API sponsored by Ensemble Data
  • SSRN version: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4676705
  • Previous title: How Does AI-Generated Voice Affect Online Video Creation? Evidence from TikTok

Major user-generated content (UGC) platforms like TikTok have introduced AI-generated voice to assist creators in complex multimodal video creation. AI voice in videos represents a novel form of partial AI assistance, where AI augments one specific modality (audio), whereas creators maintain control over other modalities (text and visuals). This study theorizes and empirically investigates the impacts of AI voice adoption on the creation, content characteristics, and consumption of videos on a video UGC platform. Using a unique dataset of 554,252 TikTok videos, we conduct multimodal analyses to detect AI voice adoption and quantify theoretically important video characteristics in different modalities. Using a stacked difference-in-differences model with propensity score matching, we find that AI voice adoption increases creators’ video production by 21.8%. While reducing audio novelty, it enhances textual and visual novelty by freeing creators’ cognitive resources. Moreover, the heterogeneity analysis reveals that AI voice boosts engagement for less-experienced creators but reduces it for experienced creators and those with established identities. We conduct additional analyses and online randomized experiments to demonstrate two key mechanisms underlying these effects: partial AI process augmentation and partial AI content substitution. This study contributes to the UGC and human-AI collaboration literature and provides practical insights for video creators and UGC platforms.

VISAGE: Designing AI Artifacts for Dynamic Self-Presentation in Matching Platforms

Kwon, Soonjae, Gene Moo Lee, Dongwon Lee, Sung-Hyuk Park (2024) “VISAGE: Designing AI Artifacts for Dynamic Self-Presentation in Matching Platforms,” Working Paper.

  • Previous title: Learning Faces to Predict Matching Probability in an Online Dating Market
  • Presentations: DS (2021), AIMLBA (2021), WITS (2021), ICIS (2022)
  • Preliminary version in ICIS 2022 Proceedings
  • Based on an industry collaboration

Online matching platforms constrain users to static profiles, producing a mismatch between the idealized self a user presents and the heterogeneous preferences of potential partners. Drawing on self-discrepancy theory, we conceptualize this mismatch as an interpersonal gap between one’s presented self and what each partner desires to see, with AI serving as a mediator to help address it. Following the computational design science perspective, we propose VISAGE, an AI system comprising two artifacts grounded in distinct human-AI collaboration principles. The augmentation artifact selects optimal images from users’ existing assets, whereas the assemblage artifact generates new images tailored to individual partner preferences. Using large-scale operational data from a major online dating platform, we evaluate VISAGE at both the user and platform levels. At the user level, model-predicted ratings suggest that both artifacts improve attractiveness ratings. Relative effectiveness varies with partner-preference heterogeneity and user impression management skill, consistent with theoretical predictions from the human-AI collaboration literature. At the platform level, agent-based simulations suggest that VISAGE can enhance matching efficiency and reduce inequality in matching opportunities, although optimal deployment strategies depend on the platform’s recommendation algorithm. The theoretical contribution of this study is to foreground the interpersonal gap as a key source of matching inefficiency and to illustrate how AI can address it at scale. The design contribution lies in actionable design knowledge, including when to deploy augmentation versus assemblage artifacts based on user and partner characteristics and how to align user-facing AI features with backend algorithmic infrastructure.