China News Service, Shanghai, November 22 (Reporter Chen Jing) The reporter learned on the 22nd that the third China Science and Technology Ethics Forum with the theme of "Technology Ethics in the Artificial Intelligence Era" was held at Fudan University. The forum attracted more than a hundred experts and scholars from various fields across the country.

At the forum, Wang Shuqin, a professor at Capital Normal University, emphasized that the instrumental value of artificial intelligence technology must always be guided by the value of human purpose. She pointed out that intelligent agents are essentially carriers of human intentions and purposes, and the responsibility for value alignment ultimately falls on humans. Scientific and technological workers should make algorithms reflect the universal value principles of human consensus.

Gong Qun, a professor at Renmin University of China, believes that the ethical risks of generative artificial intelligence are rooted in the modern conflict between instrumental rationality and value rationality. He believes that there are currently three core risks in generative artificial intelligence: the existence of black boxes brings regulatory dilemmas and ethical risks; machine hallucinations cause errors in fact and perception judgments; and the collection of personal information by platforms brings privacy risks. Faced with these challenges, Gong Qun advocates confirming a governance system with human responsibility as the core, strengthening the institutional construction of algorithm auditing, paying attention to the protection of personal data rights, and achieving the harmony and unity of technological progress and humanistic values.

Wang Chengcheng, member of the Party Leadership Group and Secretary of the Secretariat of the China Association for Science and Technology, pointed out that emerging technologies represented by artificial intelligence are accelerating and iterative, and the complexity and global nature of scientific and technological ethical challenges have become increasingly prominent. He proposed to grasp the basic principles of people-centered science and technology for good, strengthen the construction of ethical standards for artificial intelligence; accelerate the construction of the China Science and Technology Ethics Society, strengthen the supporting role of the academic community; systematically build an independent knowledge system of science and technology ethics, and enhance the international voice.

Currently, the value alignment issue of artificial intelligence has become one of the core challenges facing society. Zheng Qinghua, academician of the Chinese Academy of Engineering and secretary of the Party Committee of Tongji University, pointed out in a special report titled "Understanding and Reflection on the Alignment of Human-Machine Values" that the rapid development of artificial intelligence technology is accompanied by many risks such as algorithmic bias, artificial intelligence illusion, and intellectual property ambiguity. The fundamental crux behind this lies in the conflict between the complexity and ambiguity of human values ​​​​and the accuracy of the goals pursued by artificial intelligence. Regarding how to break the situation, Zheng Qinghua proposed that humans and artificial intelligence should be aligned in both directions: artificial intelligence needs to achieve value calibration through human feedback reinforcement learning, while humans need to adapt to the new paradigm of human-machine collaboration, and ultimately achieve human-machine value alignment. It is necessary to build a multi-dimensional and continuously evolving system to promote technology, governance and culture to work together.

At the forum, Jin Li, academician of the Chinese Academy of Sciences and president of Fudan University, said that the rapid development of emerging technologies such as artificial intelligence and brain-computer interfaces has placed high demands on the development of science and technology ethics, which is the cornerstone of civilization that balances innovation and value. Fudan University has taken the lead in establishing the Institute of Science and Technology Ethics, setting up a professional master's degree in applied ethics, and releasing the world's first AI agent system "Yijian" specifically designed for science and technology ethics review. Jinli hopes to gather more consensus through the forum, deepen the construction of the theoretical system of science and technology for good, and promote all sectors of society to pay broader and deeper attention to scientific and technological ethics issues.

Yang Min, a professor at Fudan University, introduced the Fudan University Baize large model dynamic security compliance assessment platform. This work can not only efficiently complete basic large model assessment, but also provide forward-looking decision-making support for national security governance. In terms of artificial intelligence risk warning, Yang Min revealed the potential risk of losing control of large models. He called for the establishment of an interdisciplinary governance mechanism to ensure that artificial intelligence develops on a controllable track.

Wang Guoyu, a professor at the School of Philosophy at Fudan University, and Yuan Zhenguo, a tenured professor at East China Normal University, started a themed dialogue focusing on the topic of "Reshaping Education and Ethical Boundaries in the Artificial Intelligence Era". Yuan Zhenguo proposed the three-dimensional subject interaction theory of "teacher-student-agent", pointing out that future education will break through physical walls and form a personalized full-time and space-time learning field. He said: "The essence of education in the era of artificial intelligence is to teach teachers and students to be the masters of machines in the symbiosis of man and machine."

It is reported that the forum has set up a number of special forums, focusing on human autonomy and human identity in the era of human-machine symbiosis, opportunities and challenges of artificial intelligence empowered medical and neuroscience, exploration of the possibility of artificial intelligence moral enhancement and AI empowered ethics, ethical governance of artificial intelligence and the construction of a global discourse system, methodological research on science and technology ethics, and artificial intelligence ethics in art and narrative. (over)