On the afternoon of 5 November, the 20th instalment of the "Seeking Truth and Authenticity in the Media Circle" lecture series took place at the Zhongxi Bookstore within Zhejiang University. Supported by the Zhejiang Mobile Humanities and Social Sciences Fund, the event was jointly organised by Zhejiang University's Digital Communication Research Centre, the School of Media and International Culture, and the university's "Digital Social Sciences Convergence Research Programme". This session featured distinguished speakers Professor Ge Yan, Distinguished Professor at the School of Media and Communication, Shanghai Jiao Tong University, and Assistant Researcher Zhao Hanqing from the School of Cultural and Creative Industries, Shanghai Jiao Tong University. The lecture was chaired by Professor Zhang Zike from our institute and attracted active participation from dozens of faculty members and students.
Ge Yan holds a PhD in Art Archaeology from the University of Pittsburgh. He serves as Distinguished Professor at the School of Media and Communication, Shanghai Jiao Tong University, and Distinguished Research Fellow at the Institute of Psychology and Behavioural Sciences. He also holds the position of Visiting Professor at the School of Journalism and New Media, Xi'an Jiaotong University. As a scholar long dedicated to interdisciplinary research, Ge Yan's fields encompass art archaeology, communication studies, and experimental jurisprudence. His research findings, including "Can 'Dragon' Denote 'Dragon'?" "The Boundaries of Good Deeds," and "The Influence of Cognitive Fluency on Judicial Adjudication," have been published in journals such as Social Sciences in China. Zhao Hanqing is an Assistant Research Fellow at the School of Cultural and Creative Industries at Shanghai Jiao Tong University. His primary research focus is computational social science, with particular expertise in employing simulation methods to study complex phenomena within communication studies.
The Emergence and Evolution of New Popular Literature:
A Hypothesis-Based Interpretation
Opening the lecture, Ge Yan introduced the concept of "New Popular Literature" through social phenomena such as market poets and miner novelists. First proposed in the July 2024 issue of Yanhe magazine, this concept swiftly garnered significant academic attention. The term "New Popular Arts" does not denote a specific artistic form or group label, but rather signifies a dynamic ecosystem of artistic production and consumption. Its core characteristics include: establishing a new perspective on arts and literature; the emergence of grassroots labourers as primary creators, exhibiting traits of identity fluidity; heavy reliance on digital technology, developing through formats such as short videos, online literature, and gaming; demonstrating formidable industrialisation capabilities and vast scale; and integrating diverse artistic phenomena with a "big basket" approach of inclusivity.
However, current academic research on "new popular culture" largely remains at the level of phenomenological description. Ge Yan seeks to further unravel its underlying generative and evolutionary mechanisms. To this end, he employs institutional economics theory as an analytical framework, treating "institutions" as pivotal variables shaping economic activity. He identifies four primary drivers of institutional change: the demand to reduce transaction costs; ideologies favouring fairness and efficiency; pressure to adapt to international norms; and the synergistic evolution of technology and institutions.
Tracing historical trajectories, Ge Yan charts how pivotal junctures in institutional change since the May Fourth Movement propelled transformations in the literary and artistic ecosystem. From the "literature for the common people" of 1919, to the "mass literature" of the post-1928 era, and further to the 1942 "literature and art serving the working and peasant masses" perspective following the Yan'an Forum on Literature and Art, shifts in artistic ideology have invariably been inextricably linked to adjustments in socio-political structures. Since the Reform and Opening-up, China's political institutional transformation established the principle of "centring on economic development," creating greater space for diverse artistic content. Concurrently, the economic system transitioned from a planned economy to a market economy, with cultural management systems gradually adopting market-oriented approaches.
Technological innovation became a pivotal variable following China's internet access in 1993, the steep decline in dial-up costs during the five years after 1995, and the subsequent explosive growth in user numbers. The emergence of online literary platforms such as Rongshu Xia and Jinjiang Literature City after 2000 fundamentally dismantled the monopolistic access system of traditional literary production. The creative base expanded from "institutional writers" to ordinary netizens, achieving a "creative destruction of the market access system". In 2005, the State Council explicitly liberalised the audio-visual products market, permitting private enterprises to participate in production and distribution. This further propelled industrialisation in popular music, film, and television, while emerging forms like online literature and short videos flourished through digital technology.
From 2005 to the present, market-driven artistic production has shifted from "plan-oriented" to "demand-oriented" models, gradually forming a complete "creation-distribution-consumption" industrial chain: online literature platforms incentivise intellectual property rights through paid reading mechanisms, while the gaming industry iterates products based on user data, making the industrial characteristics of artistic production increasingly pronounced. Concurrently, globalisation has accelerated the expansion of this ecosystem. Following China's accession to the WTO in 2001, competitive pressures from an open cultural market and the influx of international capital propelled the overseas expansion of Chinese online literature and mobile games, establishing them as exemplary cases of cultural globalisation. This reflects the synergistic extension of "institutional-technological-productive forces" within a globalised framework.
Ge Yan contends that new mass culture emerged within the context of reform and opening-up, driven by the interactive dynamics of institutions, technology, and productive forces, and shaped through adaptive trial-and-error exploration. The specific process can be summarised as follows: adjustments to the political system altered the perception of literature and art as weapons, reducing the political risks associated with artistic creation; economic reforms revitalised the cultural market, breaking monopolistic market access mechanisms through technological synergy and optimising the allocation of social resources; opening up to the outside world and advances in communication technology greatly enriched the content, styles, and forms of literature and art, enhancing production capacity to meet the diverse cultural and entertainment needs of the masses while creating international markets; The structural tensions of the internet era, divergent standards for evaluating artistic products, and profit-driven technological manipulation collectively constitute the evolutionary environment for new mass culture.
At the conclusion of his lecture, Ge Yan summarised that new mass culture is a product of the historical context of reform and opening-up. This exploration, marked by conflict and negotiation, can be regarded as an institutional innovation in artistic forms with Chinese characteristics. He contends that the current exploration remains preliminary and holistic, with potential for further quantitative analysis in the future.
Viewpoint Consensus and Polarisation in Social Networks:
An LLM-Augmented ABM Study
Subsequently, Zhao Hanqing presented a concrete empirical case study under the theme "Viewpoints, Consensus and Polarisation in Social Networks", systematically examining the impact of large language models (LLMs) deeply engaging in social networks upon the public opinion landscape. He noted that while opinion polarisation in the traditional social media era has drawn considerable attention, the proliferation of LLMs since late 2022 has introduced a novel research question: when AI deeply intervenes in social interactions, does public discourse converge towards consensus or polarise further? This query formed the core starting point of the present study.
To address this question, Zhao Hanqing developed the NOCAP model (Networked Opinion Consensus And Polarisation) based on classical ABM frameworks. This model simulates interactions among 50 agents across four opinion domains: economics, environment, education, and healthcare. The model design incorporates three typical network architectures—Random, Scale-Free, and Small-World networks—alongside four individual trait distributions (Uniform, Normal, Polarised, and Low Openness). These combinations yield twelve distinct conditions, each simulated over 100 steps. Two models were designed: Classic ABM and LLM ABM. The former employs only probabilistic computation without reasoning, while the latter dynamically adjusts agents' viewpoints through multi-dimensional reasoning, information integration, contextual argumentation, and human-like reasoning, with argument content generated by an LLM. Key metrics for data collection included local convergence, global polarisation, opinion entropy, and echo chamber index.
Experimental simulations revealed that LLM significantly accelerates consensus formation: under LLM-driven mode, local convergence exceeded 90% within 100 steps, with rapid convergence evident within the first 20 steps. Secondly, LLMs effectively reduce global polarisation: although both models exhibit similar polarisation levels initially, after 30 steps the LLM group's polarisation steadily declines to near-zero values, whereas the traditional model's decline is markedly more gradual. However, while LLMs serve as powerful consensus engines, they sacrifice intellectual diversity, with view entropy metrics dropping by nearly 80% under LLM-driven conditions. Across all 12 conditions, the LLM demonstrated near-universal superiority: outperforming the Classic ABM across every network architecture and exhibiting superior convergence effects across all trait distributions.
Zhao Hanqing elaborated further, stating that the primary advantage of LLMs stems from a fundamental enhancement in cognitive capabilities, achieving a leap from simple rules to complex reasoning. However, he cautioned that technological neutrality is a myth, urging awareness of the potential risks arising from the comprehensive augmentation of LLMs. These include challenges such as the erosion of viewpoint diversity, the intensification of covert echo chamber effects, and the drawbacks of rapidly formed consensus. He also outlined future research directions, including investigating the micro-mechanisms underpinning LLM reasoning quality, incorporating dynamic network architectures, and validating model conclusions through real-world social network data.
From macro to micro perspectives, and from theoretical reasoning to empirical case studies, this lecture provided an invaluable exchange of ideas for faculty and students at the School of Media and International Culture, offering profound intellectual enrichment.