YouSaid · the spoken record

Yi Zeng

lines on the record
9
first
2023-07-18
most recent
2023-07-18
sittings or episodes
1
sources
un-security-council

Every line below is reproduced as it was said and linked to the record it came from. Nothing here is summarised or generated. Directory · Search · Corrections

  1. “And we have already affirmed that a nuclear war cannot be won and must never be fought. Many countries — including but not limited to the five permanent members of the Council — have announced their own strategy and opinions regarding the use of AI for security and governance in general, and we can see there are commonalities that can serve as important inputs for international consensus, but that is still not enough. The United Nations must play a central role in setting up a framework on AI for development and governance in order to ensure global peace and security. In the context of a shared future for all, we all need to set up that agenda and framework together, leaving no one behind.”

    2023-07-18 · UN Security Council · Security Council, S/PV.9381 (2023) · source

  2. “AI is supposed to help humans solve problems, not create them. A boy once asked me if an AI-assisted nuclear bomb — apart from being a feature in science fiction — could be used to save our lives by blowing up an asteroid attacking Earth or altering its trajectory to prevent a collision with Earth. While that idea may not be scientifically solid and would be very risky at this point, it is at least about using AI to solve a problem for humankind, which is much better than empowering AI to help us attack one another with nuclear weapons on our planet, creating problems for human society and potentially posing catastrophic risks for us, the next generation or even human civilization as a whole. In my view, humans should always maintain and be responsible for final decision-making on the use of nuclear weapons.”

    2023-07-18 · UN Security Council · Security Council, S/PV.9381 (2023) · source

  3. “Considering those near-term and long-term challenges, while I am quite sure that we cannot solve the issue of AI for peace and security today, this discussion, though challenging, may nonetheless be a good starting point for Member States. In that regard, I would suggest that the Security Council consider the possibility of establishing a working group on AI for peace and security, working on near-term and long- term challenges, because at the expert level it would be more flexible and scientific to work together, and easier to reach a consensus from a scientific and technical point of view and to provide assistance and support to Council members in their decision-making. The Council should set a good example for other countries and play a significant role in this important issue.”

    2023-07-18 · UN Security Council · Security Council, S/PV.9381 (2023) · source

  4. “Both near- term and long-term AI carry a risk of causing human extinction, simply because we have not yet found a way to protect ourselves from the possibility that AI can exploit human weaknesses — and if AI were to do that, it would not even know what we mean by humans, death and life. As for the long term, we have not given super-intelligence any practical reasons why it should protect humankind, and the solution to that may take decades to find. Our preliminary research suggests that we may need to change the way we interact with one another and with other species, our ecology and the environment. That may require decision-making across the entire human species, and we will all need to work together deeply and thought-provokingly in that regard.”

    2023-07-18 · UN Security Council · Security Council, S/PV.9381 (2023) · source

  5. “I therefore want to emphasize again that AI should never, ever pretend to be human, take the place of a human or mislead humans into having wrong perceptions. We should use generative AI to assist us but never trust it to replace human decision-making. We must ensure human control of all AI-enabled weapon systems, and that human control has to be adequate, effective and responsible. For example, cognitive overload during human-AI interactions must be avoided. We must prevent the proliferation of AI- enabled weapon systems, since the related technology is very likely to be abused or used maliciously.”

    2023-07-18 · UN Security Council · Security Council, S/PV.9381 (2023) · source

  6. “For example, although the world has yet to reach a sufficiently broad consensus on lethal autonomous weapon systems, it is at least clear that AI should not be used directly to make life-and- death decisions for humans. Effective and responsible human control must be applied in order to ensure appropriate human-AI interactions. Nor should AI be used to automate diplomatic tasks, especially foreign negotiations between countries, since it is possible that it could exploit and amplify human limitations and weaknesses such as deception and distrust to create bigger or even catastrophic risks for humans. It is strange, misleading and even irresponsible that dialogue systems powered by generative AI always use expressions such as “I think” and “I suggest” in their arguments when there is no “I” or even “me” in AI models.”

    2023-07-18 · UN Security Council · Security Council, S/PV.9381 (2023) · source

  7. “AI should be used to connect people and cultures rather than disconnecting them. That is why we created an AI-enabled culture-interaction engine that finds commonalities and diversities among various UNESCO World Heritage Sites. Those World Heritage Sites have shown us that we are not so divided in terms of culture as we may think, and the commonalities serve as roots and hands that reach out, helping us to appreciate, understand and even learn from the diversities in various cultures. The current examples of AI, including, more recently, generative AI, are all information-processing tools that seem intelligent but have no real understanding and are therefore not truly intelligent. That is of course why they cannot be trusted as responsible agents to help humans make decisions.”

    2023-07-18 · UN Security Council · Security Council, S/PV.9381 (2023) · source

  8. “However, I think those are topics of fundamental importance for the future of humankind, and Governments should definitely work together to address them. Military AI and AI for peace and security, while closely related, are fundamentally different in important aspects. We should advance the use of AI for international peace as an essential pillar of the Sustainable Development Goals, with a view to reducing security and safety risks rather than increasing them. When we are thinking about AI for good from a perspective of peace and security, rather than seeking ways to create disinformation for military and political purposes, it would be much better to work on using AI to identify disinformation and misunderstandings among different countries and political bodies, and to use it not to attack networks but to defend them.”

    2023-07-18 · UN Security Council · Security Council, S/PV.9381 (2023) · source

  9. “Mr. Yi Zeng: My name is Yi Zeng. I would like to take this opportunity to share with the Security Council my personal view of artificial intelligence (AI) as a force for good in terms of international peace and security. I hope it will be useful in promoting the discussion and understanding of the necessity of global governance for AI. There can be no doubt that AI is a powerful enabling technology for advancing global sustainable development. When investigating the use of AI for the Sustainable Development Goals (SDGs), we have found that most of the efforts focus on AI for quality education and for health care, while very few efforts have been put forward on many other important topics such as AI for biodiversity, climate action or peace.”

    2023-07-18 · UN Security Council · Security Council, S/PV.9381 (2023) · source