Gene­ra­tive AI's Norma­tive Rati­o­na­lity

This lecture examines how generative AI models produce specific forms of normativity, assign a distinctive position to the user, and claim an epistemic authority that emerges from what I describe as a “gaze from everywhere.” It argues that these normative operations are inseparable from generative AI’s political economy. At the current stage of capital accumulation, AI systems depend upon and reinforce the concentration of wealth and power in the hands of a very few, at the expense of the majority. This concentration of power shapes both the values embedded in models through post-training and their alignment with state interests in the race for AI supremacy between the United States and China. Generative AI’s environmental harms, substantial energy demands, and financial barriers to accessing the most advanced models impose unequal burdens, with disenfranchised populations and the Global South bearing the brunt of the negative impacts. In this context, the university must confront the question of its own autonomy and educational purpose: will it relinquish its autonomy to generative AI and train students to become data annotators, evaluators, and improvers of generated content, thereby subjecting them to AI’s normative rationality, or will it cultivate their capacity to act as autonomous and responsible social subjects?

• Vertiefender Workshop
• Vortragende