Big data drives changes in research paradigms across disciplines
Opening ceremony of the stragetic seminar Photo: CAAI
The Chinese Association for Artificial Intelligence (CAAI) recently hosted the Interdisciplinary Strategy Seminar on Social Computing and Social Intelligence in Luoyang, Henan Province. The event gathered scholars from Peking University, Tsinghua University, the Chinese Academy of Social Sciences, Renmin University of China, the Chinese Academy of Sciences, the China Population and Development Research Center, and other institutions. Discussing major issues in social computing and social intelligence, attendees formed scholarly ideas that will build future research developments.
Disciplinary crossovers
Scholars at the seminar agreed that the social sphere, with people at the core, is a highly sophisticated system, calling upon interdisciplinary research in the natural sciences, social sciences, computing science, biological science, and other fields.
Meng Xiaofeng, director of the Social Computing and Social Intelligence Committee at CAAI believes that big data gave birth to social computing. In a post-industrial society, it becomes increasingly difficult to predict and simulate complex social issues. Nevertheless, big data and artificial intelligence lend new materials and approaches to mapping, analyzing, and foreseeing them, opening up opportunities for social computing. Meanwhile, social computing and social intelligence can set an example in terms of disciplinary crossover.
Zhou Xiaolin, a professor from the School of Psychological and Cognitive Sciences at Peking University, suggested that computational modeling sets the tone for social cognition research. In this regard, computational modeling integrates artificial intelligence and cognitive sciences, so that they interweave. The social cognition arena can use the latest AI technical frameworks and data analysis methods to refine and investigate complex human behaviors, thus quantifying behaviors into an operable algorithm. AI study can adopt these algorithms to understand people’s cognitive computing characteristics when dealing with complex social situations, and further develop a more versatile and anthropomorphic machine learning framework, driving fundamental reforms to the traditional social computing field. Therefore, AI algorithms, based on the characteristics of social cognition, require a deep integration of AI and cognitive science.
Lyu Xin, a professor from the School of System Engineering at the National University of Defense Technology, introduced a cutting-edge study of higher-order dependencies for complex systems. He said that most indicators and models in theoretical systems of existing network sciences are based on Markov Assumptions. However, networks in reality are multi-dimensional, meaning that we can’t measure interactions between nodes by simply accumulating their connections’ transmission effects. Therefore, higher-order network dependencies deserve more attention. We should use methods other than Markov models to do a network study.
Big data’s driving force
Seminar attendees are confident that China is entering an Industry 4.0 stage, which is mainly characterized by social intelligence and the digital interconnection of all things. Data accumulation has reached a breakthrough in terms of depth and breadth, and data infrastructure is expanding its role in society. Experts and scholars are tasked with offering solutions which integrate social science research with big data and AI technology, then surpass traditional systems to discover new phenomena based on the materials and structures of digital society. Thus, researchers can explore new research topics and establish new methodologies.
Wu Gang, from the Department of Management Science at the National Natural Science Foundation of China, spoke on big data’s impact on business, finance, healthcare, public management, and other fields. He believes that a management decision-making paradigm, driven by big data, has taken shape. Wu introduced the “PAGE” management decision-making method (Paradigm, Analytics, Governance, and Enabling), which touches upon four dimensions of big data’s influence, namely a paradigm shift, innovation methods, new governance, and value creation.
Luo Jiade, a professor from the School of Social Sciences at Tsinghua University, considers social computing a big data research method based on social science theories. He advanced three social computing networks to connect decisions and actions. The first is human decisions and actions + social networks (internet and mobile communications); The second is AI “decisions” + Internet of Things (IoT); The third is human-machine interaction + human-machine co-construction networks.
Liang Yucheng, a professor from the School of Sociology and Anthropology at Sun Yat-sen University said that the development of computational social science must solve the escalating problem of authentic data scarcity after digitization. By updating cognitive models, we can use absolutely scarce data in quantity and relatively scarce data in authenticity to understand the world we live in. Both the Agent-Based Modeling (ABM) paradigm and the machine learning paradigm have presented directions for potential breakthroughs.
Qi Jiayin, a professor from the School of Management at Shanghai University of International Business and Economics, introduced the fifth research paradigm (following experimental science, theoretical science, computing science, and data-intensive science) and her thoughts on cross-disciplinary innovation. She comprehensively summarized the fifth research paradigm’s current situation, both domestically and internationally, and contended that the field is still in its infancy.
Qi also analyzed open-source platform management theories and related key technologies and proposed three future research focuses, namely the intellectual property strategy, a basic theory on open-source strategy, and a system for concepts, spirituality, and culture. From open-source software to open-source hardware and open-source commerce, open-source concerns both technological innovation and commercial applications, behind which an inter-disciplinary group may generate breakthroughs in the theoretical innovation of economics, management, sociology, and law.
New discoveries
Experts agree that it behooves the social computing field to discover new materials in the context of the Internet of Everything (IoE), to discuss new problems in its application, and to explore new methods.
Wang Jingyuan, an associate professor from the School of Computer Science and Engineering at Beihang University, introduced predictive methods for interpretable deep learning that integrates ideas in econometrics and analyzes the conceptual framework of urban data intelligence. He believes that future urban safety problems may come from AI’s interpretability, and he emphasized the importance of interpretable prediction, introducing interpretable prediction through traffic speed prediction, fuzzy cognitive maps, and other applications.
Huang Kuangshi, deputy director of the Social Computing and Social Intelligence Committee at CAAI, discussed human development, challenges, and reconstruction in the AI era. He said that human biological attributes, consumption attributes, social attributes, and management decision attributes will make great strides in the AI era. In particular, study of biological attributes, aspects such as gender, age, health, and fertility, is already imminent or underway. However, in the process of expansion, society faces problems such as low fertility rates, the loss of human autonomy, and the enhancement of machine autonomy. Therefore, in the AI era, population concepts need to be reconstructed, as does the relationship between humans and AI.
Academic community
Seminar participants held heated discussions on social computing and social intelligence, and came to an academic agreement titled “Social Computing and Social Intelligence: Challenges and Opportunities.” This consensus summarizes eight major challenges in social computing from three perspectives: data governance challenges in social computing, interactions between AI and human society, and the research and development of complex social systems.
Major challenges include data ethics and data governance, open-source platform building and management, human and social reconstruction, cognition-based AI, complex system modeling and analysis, simulation system construction and research, causal reasoning and causal discovery, and global public crisis management. Attendees called for a paradigm shift in scientific research, to create a solid base which will transform challenges into opportunities.
Participants believed that with the continuous development of the IoT, 5G, and other technologies, intelligent IoE and other information technologies will describe society more richly and concretely through sharpened data collection methods, pushing social science researchers to understand people and society from more diverse perspectives.
Social computing should shine more light on humans and human nature, and the current human-centric AI orientation should adopt clearer humanity-centric AI concepts in development, said Huang. Meng advocated for the fields of social computing and social intelligence to commit to creating an academic community built on benevolence, integrity, and compassion.
Meng Xiaofeng, Huang Kuangshi and Ma Youzhong are from the Social Computing and Social Intelligence Committee at CAAI.
Edited by MA YUHONG