Challenges and paths to responsible development of generative AI
Deng Zhenyu
Centre for Studies of Intellectual Property Rights, Zhongnan University of Economics and Law
Abstract： Generative AI is driving a new round of changes in the science and technology industry, but it also triggers many governance risks, such as data misuse, poor quality of training data, algorithmic black box and algorithmic discrimination are common, but the current regulatory concepts and measures are still lagging behind. In order to achieve the responsible development of generative AI, on the governance path, it is necessary for multiple subjects such as users, developers and regulators to participate in the governance in a coordinated manner from multiple dimensions such as technology and law, to build a data quality management system and establish an open and shared public training data platform. Algorithmic filing and registration, review and assessment mechanisms are implemented, to achieve fair and impartial and agile governance of algorithms. The system of attribution of responsibility for generative AI infringement is improved, and the application of regulatory sandbox system is explored, so as to realise the responsible development of technological innovation.
Key words : responsible development; generative artificial intelligence; open sharing; regulatory sandbox
