Home: Home > News > Content


Advancing AI trustworthiness: From Principles to Practices

June 2, 2022

The BAAI Conference 2022 - AI Ethics and Governance Forum, hosted by the Beijing Academy of Artificial Intelligence (BAAI), with the Institute for AI International Governance (I-AIIG) of Tsinghua University and the Center for AI Governance as the co-organizer, opened on June 1, 2022. The event was moderated by LIANG Zheng, vice dean of I-AIIG.

The rapidly growing capabilities and an increasing presence of AI-based systems in our lives, leave largely unanswered a broad range of important short- and long-term questions related to the social impact, governance, and ethical implications of these technologies and practices.

“According to incomplete statistics, in recent years, more than 100 documents related to AI ethics and governance have been released around the world,” said ZHAO Zhiyun, Dean of New generation Artificial Intelligence Development Research Center, Ministry of Science and Technology. “Though the regulatory approaches are diverse, a common understanding has been reached among multiple stakeholders. We should get the most out of AI while still protecting important human values.”

China, the United States, and the European Union are now emerging as the main competitors for global leadership in AI, and AI regulation is growing to be one of the key parts of this competition.

Comparing the laws and regulations of AI governance in China, the U.S., and the EU, Wendell Wallach, an Academic Member of I-AIIG, pointed out that China’s AI regulations were moving fast, running some large-scale experiments that the rest of the world could watch and learn something from. In the U.S., the National AI Initiative Act of 2020 became law in January 2021, accelerating AI research, development, and innovation, while advancing trustworthy AI. With quick reactions, the EU unveiled the General Data Protection Regulation (GDPR) and Artificial intelligence act in succession, and proposed an EU regulatory framework on AI. “Each approach has its pros and cons, as AI governance is complicated, one approach may not be able to tackle all. We need to work together in defining which will benefit the many.”, said Wendell.

CHEN Xiaoping, Professor of the School of Computer Science and Technology, University of Science and Technology of China, drew attention to the main challenges of AI governance -  rationality, controllability, and social impact. “It is noteworthy that these challenges require approaches respectively, and the role of the law, ethics, and technology in governing AI systems are thus more relevant than ever before.”, CHEN explained.

One other fact that cannot be ignored is the algorithmic bias, which can continue to plague AI technology and governance. However, there is less discussion of what bias is. Without fully understanding, it will be challenging to solve.

Osamu Sakura, Professor of the Interfaculty Initiative in Information Studies of the Tokyo University, and his colleague Haruka Maeda conducted a survey using a questionnaire to reveal how Japanese people feel about cases of algorithmic bias. “The result helps us consider and propose an ethical education program for users, designers, and programmers. All of them should receive training on how to recognize and avoid bias in AI.”

As AI governance is complicated, researchers, policymakers, and practitioners are still exploring the best roadmap to guide AI governance, and the voice for building a new type of governance pattern is rising.

“A new and innovative AI governance mechanism should be explored,” reflected LI Renhan, Chief Adviser of Artificial Intelligence Institute at Shanghai Jiao Tong University. He proposed that the AI evaluation system could help to balance technological innovation and governance in the development process. “It is an important infrastructure for future AI governance, and can contribute to building a robust, responsible, and integrated AI governance mechanism.”

Panelists touched on a variety of opportunities that might unlock AI governance’s potential. XUE Lan, dean of I-AIIG, summarized and gave five suggestions for future AI governance: Facilitate building an AI governance mechanism, Promote the implementation of laws, regulations, and standards, Construct an AI evaluation, early warning, and control system, Enhance AI ethics and governance education, and Strengthen international communication and cooperation.

“AI governance is an interaction and construction process among multiple stakeholders, so the communication is particularly important to eliminate information asymmetry.”, XUE concluded.

上一条:Tsinghua ramps up efforts for developing sustainable digital ecosystems in China and Indonesia 下一条:Symposium on AI and Climate Action kicks off at Tsinghua University