Social Sciences in China (Chinese Edition)

No. 8, 2024

 

Reconceptualizing Agency with Generative Artificial Intelligence

(Abstract)

 

Yin Jie

 

Generative Artificial Intelligence, in its interaction with humans, demonstrates an autonomous interaction capability that is distinctly different from previous forms of artificial intelligence, thus exhibiting the characteristics of agency and forming a new pattern of human-machine interaction. This challenges the traditional concept of agency. From an interactive perspective, the agency of biological organisms is defined the interactions between the individual and the environment, driven by self-organization. Human agency is defined by the individual’s interaction with the natural and socio-cultural environment, driven by emotions, language, and culture. The agency exhibited by Generative Artificial Intelligence in its interaction with humans is defined by the interaction capability driven by language generation. Agency can thus be viewed as the potential action capacity of the entities participating in the interaction, determined by the driving forces of the interaction. This form of interactive agency is not derived from the concept of the agent in the traditional sense, nor does it necessitate that all participants in the interaction possess intentionality. Instead, interactive agency encompasses the goal-directedness and autonomy of the entities engaged in the interaction, representing a new form of artificial agency and providing new semantics to the concept of agency.