"Artificial intelligence helps society achieve well-being and changes the human track through ultimate automation, but this track may not be the direction we preset. Some rapidly disruptive technologies have a huge impact on human decision-making, which has led to AI ethics and supervision The issue has become very critical. "The director of the European Center of Taihe Think Tank, said at the seminar on building an artificial intelligence ethics and governance network held on May 22.

As a higher-level computer program, artificial intelligence is increasingly permeating all aspects of human life. While having bright hopes for the future of AI, human beings are also worried about this new thing. As countries continue to seek to maximize the benefits of artificial intelligence, how to deal with their ethical and regulatory issues to minimize systemic risks and negative effects is also hotly debated.

"When we talk about ethics, we always talk about minimizing risks." Wendell Wallach, director of the science and ethics discipline of the Center for Bioethics at Yale University, emphasized the goals pursued by AI ethics.

Photo provided by the organizer

What AI principles do humans need?

In March this year, the China New Generation Artificial Intelligence Governance Professional Committee held its first meeting. As members of this organization, Xue Lan and Zeng Yi also participated in the seminar hosted by Taihe Think Tank. Xue Lan, director and professor at Tsinghua University's Su Shimin Academy, believes that China often has great enthusiasm for innovation. China actually missed the first three industrial revolutions. In the era of artificial intelligence transformation, China must take risks together and explore the principles of AI. .

"Unified principles or standards of AI are unnecessary and impossible, because no one knows what stage the future of AI can develop to." Zeng Yi, deputy director of the Brain-like Intelligence Research Center of the Institute of Automation, Chinese Academy of Sciences, said. Now about 42 different AI principles have been proposed, and it can even be reduced to 28. But through analysis, we can see that in fact, there is no AI principle that can cover all content.

Last month, the EU issued guidelines on AI ethics, which should follow seven principles when developing AI in the future, calling for "trustworthy artificial intelligence." Specifically, it includes human role and supervision, technology robustness and security, privacy and data management, etc.

When the public is talking about AI governance, words such as security, ethics, ethics, and privacy are repeatedly mentioned. In the view of some researchers, after the arrival of the era of artificial intelligence, how to improve the transparency of artificial intelligence should be put first. Wendell Wallach is part of this view, saying that “we are creating increasingly complex systems now that may exceed our capabilities.” Artificial intelligence is like a black box, and the most urgent task is to do as much as possible Understand it.

In his speech, Pan Tianyou, deputy director of Microsoft Research Institute of Asia, focused on "people". He believes that artificial intelligence is ultimately a tool for human beings, and that it is beautiful at the beginning of creation, and that the governance of artificial intelligence will eventually return to people themselves. Pan Tianyou also advocates that artificial intelligence should not be enjoyed by a small number of people, and that artificial intelligence technology should replace repetitive labor and release manpower.

Consensus building requires global collaboration

It was pointed out at the seminar that climate change is a challenge facing the world at present, and the development of artificial intelligence also has this trend. It is no longer a problem of a certain region or country, but requires global collaboration to establish a widely recognized one. AI principles.

Faced with ethical problems that artificial intelligence has already or may cause, outside the national level, technology giants have set up ethical committees to meet challenges. "Daily Economic News" reporter noticed that Amazon, Microsoft, Google, IBM and previously jointly established a non-profit artificial intelligence cooperative organization (on AI), which Apple joined in 2017, and various companies also have their own self-reliance. AI Ethics Committee.

Earlier this year, the Beijing News reported that the Google AI Ethics Committee, which had just been established for a week, was actually disbanded. Wu Dekai, professor of the Department of Computer Science and Engineering of the Hong Kong University of Science and Technology, is one of eight members of Google's organization. At this seminar, Wu Dekai clarified, "Google has never used the term committee. It is not governance or consultation. There is this Spread misunderstandings.

"In AI, we need to conduct in-depth translation, not very shallow translation. We need to describe without any prejudice or position, and only after we understand each other can we learn from each other." Wu Dekai gave an example and emphasized that we must hear Different cultures, different voices.

Liu Yang is a researcher from the Center for Future Intelligence at Cambridge University, and his other identity is a researcher in the Department of Philosophy. "China cannot ignore the power of some innovations in artificial intelligence, but in the work it is engaged in, I realize that many reports have some misunderstandings about the development of China's artificial intelligence, and I hope that more Chinese voices will be brought to the West."

Wang Yiwei, a professor at the School of International Relations at Renmin University of China, said in the discussion that the "rich and poor gap" in artificial intelligence may increase in the future, and humans will rely more on intelligence in the future, which intensifies the solidification of social classes. On the other hand, Chinese culture emphasizes harmony and harmony, and connectivity and openness, and the new generation of artificial intelligence technologies are very easy to achieve such goals.

Daily Economic News