The scoping review examines how existing research explains social impacts and risks of AI, offering insights into the development of IA modernization in the AI era.
Artificial intelligence (AI) is being integrated into impact assessment (IA) practices increasingly, offering new opportunities for the modernization of IA with its technological potential. As AI becomes a new tool for IA, practitioners are facing more complex scenarios when selecting and using assessment tools, which places particular demands on their understanding of the social impacts and risks involved. Guided by this awareness, we conduct a scoping review of relevant articles in English published over the past decade, with the aim of exploring how existing research conceptualizes and explains social impacts and risks of AI. The review reveals a complex relationship between the generation pathways and social manifestations of the social risks associated with AI. Different pathways may lead to similar social consequences, while comparable pathways may produce divergent social impacts within different conditions. Such a relationship depends not only on specific technical characteristics but also on the manner in which social factors contribute to the risk formation process and interact with one another. However, while existing research has done a relatively thorough work in identifying the negative outcomes caused by AI, the components and mechanisms involved remain to be further studied. The review provides a reflective perspective on IA in AI era, emphasizing that modernizing responsible IA involves not only the introduction of new tools but also the prudent selection and use based on a comprehensive understanding of the underlying risk logic.
Weichuan Kong is a Ph.D. student from School of International Affairs and Public Administration at Ocean University of China, with a research focus on Social License to Operate.