인공지능은 어떻게 윤리적 ‘대상’이 되는가? 법적 모형들의 수행성과 반수행성에 관한 분석

Amidst highly divergent perspectives on how similar generative AI is to humans in terms of capabilities, functions, and agency, this technological system, which processes vast amounts of information swiftly and generates new outputs based on that data, is rapidly evolving under the dominant influence of a few large tech corporations. This fact in itself poses significant challenges to the legal and ethical principles that regulate information, media, and knowledge, leading to the introduction of AI-related legislation and policies worldwide. This study seeks to analyze the emergence of fractures in existing legal models, as observed in these legal discussions, and the dynamics by which regulatory gaps are being filled through the reconstruction of ethical frameworks via new moral principles, business models, and technological innovation—through the lens of “performativity and counterperformativity of legal models.” By utilizing Actor-Network Theory (ANT), which emphasizes the materiality of performance beyond mere performativity, and the concept of counterperformativity, which focuses on model reconstruction resulting from performative failure or malfunction, this paper examines conflicting texts related to major AI policy issues: monopolization of AI technology, intellectual property rights over training data, and the allocation of responsibility for AI system failures. Through this, the study aims to elucidate how legal discussions surrounding AI are closely intertwined with fundamental issues in the philosophy of technology