1. Will AI take away jobs from humans in the future?

2. How can we improve the legal governance of AI?

3. What are the similarities and differences between the U.S. and China in models of AI governance?

On the morning of July 5, the 2024 World Artificial Intelligence Conference (WAIC 2024) - Intelligent Society Forum, organized by Tongji University and the Yangpu District People's Government, saw keynote speeches kick off. Six distinguished guests from various fields both at home and abroad, including AI, public administration, law, cybersecurity and economics, were invited to deliver speeches. These leading experts provided an elaborate and compelling exploration of the evolution of AI, from its origins to the present and into the future.

Jin Yaochu: Could AI be a "Clever Hans"?

Jin Yaochu, Member of the European Academy of Sciences, IEEE Fellow and Chair Professor for AI, Westlake University, as the first keynote speaker, spoke under the title of "A Glimpse into AI Governance". He briefly introduced the development history of AI, elaborating on the risks it faces, technical remedies and government management.

He mentioned the three "springs" and two "winters" of AI since 1956, with the most impressive being the third "spring" brought about by Chat GPT. Its dialogue fluency is astonishing and has propelled AI large models into a stage of rapid development.

On the risks faced by AI, Jin outlined six critical elements associated with AI: security, privacy, equity, interpretability, robustness and responsibility. For each element, he raised high-risk issues and proposed technical solutions, provoking thought among the attendees. Speaking of the interpretability of AI, he illustrated his point with the historical case of Clever Hans, a horse in the early 20th century that amazed audiences by performing basic arithmetic after a period of training. It could accurately tap out answers with its roof. However, it was later revealed that Hans was not calculating but responding to the subtle cues in the questioners' facial expressions. "Could AI be a 'Clever Hans'? Perhaps causal learning can help overcome this deficiency."

Finally, on the issue of government management of AI, Jin shared his views on "What should we actually regulate?" He believed that the government should regulate AI products and application scenarios, rather than the technology itself.

Wu Zhiqiang: "Humans and AI should coexist in harmony."

Wu Zhiqiang, Member of the Chinese Academy of Engineering and former Vice President of Tongji University, spoke under the title of "Social Intelligence on the Breakthrough of HAI." HAI, or Human AI, integrates the concept of "human" with AI to achieve deep cooperation between humans and AI, aiming for a profound collaboration that fosters a symbiosis between human intelligence and AI.

For Wu, the philosophy of HAI is founded on five core principles: human-centric cooperation, transparency and mutual trust, security and mutual safeguarding, controllable interaction, and the joint cultivation of ethics. "The development of AI must be centered on human needs; this is the fundamental starting point," he said. "For instance, when planning a city's functional layout, we conducted a big data survey involving hundreds of thousands of residents to find out what their happiness comes from and what they complain about most, thereby guiding our decision making. I remember that in Xiamen, locals are most happy about 'satay noodles', so we must not forget the special noodles when planning.

In Shanghai, the philosophy of HAI is referred to as HAI-ism. Wu said that HAI represents the interaction between humans and AI, a synergy that continuously drives the intelligent society forward, embodying a human-centric approach where AI is harnessed for benevolent purposes.

Gabriele Mazzini: An Insight into the EU Artificial Intelligence Act

Gabriele Mazzini, an official of the European Commission and the lead author of the proposal on the Artificial Intelligence Act by the European Commission, presented a comprehensive overview via video link, detailing the background, nature, content, and application scope of the act.

In May 2024, the Council of the European Union officially approved the EU Artificial Intelligence Act, marking a milestone as the world's first integrated legislative framework for AI governance. The act categorizes AI systems and sets corresponding regulatory requirements based on levels of risks, instituting a universal and holistic approach to AI regulation that covers data management, transparency, interpretability and human supervision. Following the adoption of the act, the EU declared the establishment of an AI office to oversee AI development in accordance with law.

Mazzini emphasized the term "risk". He explained that the Artificial Intelligence Act adopts a "risk-based" approach, classifying AI systems into four categories according to their potential impact on users and the society: unacceptable risks, high risks, limited risks and minimal risks. The higher risk levels are, the stricter the control is. The act places particular emphasis on the regulation and constraints of high-risk AI systems.

At the end of his speech, Mazzini reminded that: "After the Artificial Intelligence Act comes into effect, not all rules will take effect at the same time, but will become effective by phase in according to their specific content."

Matt Sheehan: Exploring New Paths for Sino-US AI Governance

Matt Sheehan, a researcher at the Carnegie Endowment for International Peace, began his speech titled "US-China AI Dialogue: Obstacles and Way Forward" by referencing the 1st Sino-US Intergovernmental Dialogue on AI that took place in Switzerland in May. He recounted the dynamic exchanges and fruitful outcomes regarding cutting-edge AI governance between the two countries, and envisioning the interactive development of AI among policymakers, scientists and researchers from both countries.

AI, as a revolutionary technology, is experiencing a surge in growth. At the 1st Sino-US Intergovernmental Dialogue on AI, the two sides engaged in profound, specialized and constructive discussions on the risks of AI technology, global governance and other issues of concerns. "This is an important dialogue because addressing the issues at the forefront of AI requires the joint efforts of China and the U.S.," said Sheehan. "China and the U.S. need to work together to address the risks posed by AI, extending beyond governmental levels to foster more effective dialogues among experts, scholars, scientists and businesses, which I believe is vital for ensuring the security of AI between our two countries."

Sheehan maintained an optimistic outlook on the future of AI development for both China and the U.S. "Professionals from both countries can proactively engage in non-governmental dialogues to share insights, learn from each other and exchange effective AI practices, thereby continuously driving AI forward."

Karman Lucero: The "Galactic Collision" between China and the U.S.

Karman Lucero, a research fellow at the Paul Tsai China Center, Yale Law School, delivered a keynote speech titled "Challenges of AI Governance between China and the U.S."

In his speech, Lucero introduced the AI governance model of the U.S. and its engagement in international AI governance. He also provided insights into China's AI governance ecosystem, comparing the two systems from an institutional perspective. He noted that while the AI governance systems of China and the U.S. differ, there are also some similarities in certain aspects.

"The AI of China and the U.S. are like two cosmic galaxies," said Lucero. "The collision of these galaxies isn't about destruction; it's about a mutual transformation that could illuminate and enhance both."

Su Jun: Will the Development of AI Affect Human Employment?

Su Jun, Dean of the Institute for AI International Governance, Tsinghua University and Head of the National Expert Group on AI Social Experiments, delivered a speech titled "Responding to Technological Transformations in the Intelligent Society: Historical Experience and Current Challenges". He explored the potential and existing social problems against the backdrop of the intelligent society from four aspects: AI and new quality productive forces, pressing issues in intelligent society governance, historical experience in responding to technological transformations, and building a humanistic intelligent society. He reflected on past experience and practices, and shared a wonderful vision for the future humanistic intelligent society.

Group polarization caused by information cocoons, flexible employment and labor substitution, the evaluation of "values" in large AI models, AI and energy issues, online games and cyber civilization...countless governance issues have arisen in the human intelligent society. Regarding employment, Su suggested that the rapid development of AI may lead to a third category of employment beyond employment and unemployment: freelancing. "As AI assumes human tasks and the demand for labor in social roles decreases, some people may withdraw from productive labor or cease to rely on labor for economic gain. This is freelancing. Humans will need to address the question of 'how to spend meaningless time' in this situation."

In response, Su presented three historical approaches to dealing with technological transformations. Firstly, social elite, especially intellectuals, driven by a humanistic spirit and a strong sense of social responsibility, pay close attention to the social impact of groundbreaking technologies. Secondly, the government, prioritizing public interest and shared values, employs a variety of policy instruments to steer technological innovation and manage risks effectively. Lastly, the public enhances their scientific literacy and heightens their risk awareness, actively engaging in the governance of technological transformations.

"What kind of intelligent society do we expect to build for future generations?" Su asked. "A society that is humanistic and intelligent. In this society, science and technology are highly developed, the application of intelligent technologies is pervasive, there is a harmonious balance between technical rationality and ethical values, and a state of symbiosis is achieved among humans, the environment and technology..."