主题：Will Human Creatives Survive Generative AI Shocks?

主讲人：Jiasun Li, Professor, George Mason University

时间：4月1日（周三）上午10:00-11:30

地点：4-101教室

语言：英文

摘要：

Concerns are mounting about the impact of generative AI on human jobs: With ongoing improvements, would generative AI models eventually replace all human creatives one day? What kind of tasks may resist generative AI shocks, if any? What appropriate actions should human creatives and policymakers take in response to generative AI shocks? This paper develops a flexible modeling framework to shed light on these questions and synthesize empirical evidence. First, in a stylized static model inspired by familiar insights on the impossibility of an informationally efficient financial market [e.g., Grossman and Stiglitz 1980], we demonstrate robust human–AI coexistence, that is, even if existing generative AI models' capabilities reach their theoretical limit, they still cannot fully replace human creatives. This is because generative AI has to be trained on some existing content. Hence, if humans are (hypothetically) fully replaced by generative AI and stop producing contents that capture new happenings in the physical world, then generative AI can only be trained on stale content and thus offer low values, creating rooms for human to re-enter. The static model can be microfounded by a dynamic renewal model and readily enriched with preference/skill heterogeneity or task complementarity to identify which human tasks will sustain in the generative-AI era. Overall, we provide a unified framework to analyze multiple aspects of human–generative AI interactions and labor market implications, with clean comparative statics and testable predictions.