The Triple Dilemmas of Technical Governance of Generative AI and Responses

With the advent of the era of digital intelligence, the proliferation of generative AI applications has aroused great concern regarding ethical dilemmas and risks. Fairness, accuracy, accountability, safety and human supervision are core principals of Responsible AI. However, the technical governance of generative AI has triple dilemmas: the ontological tension among values, representational biases in training data, and the passive protection system. This also poses actual challenges to how generative AI can be "responsible". Under the historical context of globalized risks and intensified international competition, accelerating the establishment of cooperative mechanisms for global governance, practicing the governance wisdom of cultural inclusion, and returning to the epistemological orientation towards the root causes of social problems are the keys to realizing the "responsible" principles in the face of the potential ethical and social risks of generative AI.