The following are the outputs of the captioning taken during an IGF intervention. Although it is largely accurate, in some cases it may be incomplete or inaccurate due to inaudible passages or transcription errors. It is posted as an aid, but should not be treated as an authoritative record.
***
>> Hello? Hello? Is that okay?
>> Hello, everyone. Please wear your headphone. Hello, everyone. Please wear your headphone.
Okay. Good afternoon, distinguished guests, esteemed colleagues, and friends. Ladies and gentlemen, it's my great honor to welcome all of you to this session. On behalf of the organizing committee, I would like to express my deep respects and gratitude to all who have been contributing to this important discussion
Today, we are gathered at pivotal moment in history. The artificial intelligence is driving huge transformation toward intelligence new frontiers, technological breakthroughs, and innovative models. This transformation brings enormous opportunities for development across all sectors. However, it also introduces a range of complex challenges, including concerns, social inequity, privacy, and security risks.
This session aims to address these pressing issues from a global perspective, and aligning the latest trends, major challenges, and future opportunities in the development of intelligent society. Through the lens of government and international collaboration, we will explore these environmental and adaptive approaches needed to navigate the governance transition of intelligent society
We are privileged to have exceptional panel of speakers, each of whom bring unique expertise and global perspectives to this discussion. Now, allow me to introduce them. Gong Ke, former President of the world of federation of engineering organization, Executive Director of Chinese Institute of New Generation Artificial Tj Development koim strategies and former President of Nankai University
Mr. oxford university and director of multilateral AI
Professor the Harvard Institute of technology and editor and chief of journal of public administration. Kevin Desouza, Professor of business technology and strategy at School of Business Queensland University of Technology. Ru Peng, pro pesser of school of prick policy and management, Tsinghua University
And from Gambia in Africa
To begin the session. It is my distinct pleasure to introduce our first speaker, Gong Ke who will deliver keynote address in building intelligent society. Please join me in giving warm round of applause to welcome Gong Ke to the stage. Thank you
>> KE GONG: Thank you so much for the introduction. I will take this opportunity to introduce you briefly about the Chinese national plan for building the intelligent society. And you may already know that the Chinese government has released a top tier plan to report the new generation of artificial intelligence development from 2017 to 2030. That's a long term high level plan. And this plan is in China as a 1, 2, 3, 4 planning. So 1 means to set up one national, open, and collaborative AI technological innovation system. One nationwide system. 2 means to master the two attributes of artificial intelligence. One is the technical feature, another one is social feature of artificial intelligence. Three means three in one promotions. That means to advance the technical R&D, the production and manufacturing and industrial nurturing in three in one manner.
The 4 means 4 packs to be supported by the AI environment that are STI environment, Science, Technology, and innovation development. The economic growth and social progress and the national security.
So within these frameworks, the plan identifies six pivotal tasks with building an intelligent society alongside with fostering technological innovation system, nurturing intelligent economy, and enhancing digital infrastructure.
So the goals for building an intelligent society is that, in one word, it's the building of safe and convenient intelligent society. And the plan outlines some objectives. First is to accelerate the penetration of AI to to elevate the equality of life and create an omni present intelligent environment, significantly enhancing the social services and social management.
The second is delegate being simplistic, repetitive, and hazardous tasks to AI, thereby, fostering human creativity and generating comfortable employment opportunities.
The third objective is to diversifying and enriching on demand intelligent services to maximize accessibility to high quality social services and convenient lifestyle. The fourth objective is elevating the standard of intelligent, social governance, rendering societal operations safer and more efficient.
So under these goals and objectives, there are some tasks for building intelligent society outlined and highlighted in the plan. It's the efficient and intelligent process (audio is breaking up).
(captioner cannot understand audio).
We have adopted a new law in China two years ago to protect the private information. Then safety and ethical norm policies, establishing the AI safety regulation and social impact, infused with ethical guidelines released by the Chinese Government three years ago. Finally, regional development policies to encourage regional and localize the development initiatives.
So with these policy frameworks, China has got, I think, notable progress in building an intelligent society. For example, developing application scenarios in social services in healthcare, China the artificial intelligence image screening technology has been dramatically improved, pre diagnostic and diagnostic of critical illness, such as cancer.
In education intelligent systems has optimized the resource allocation for equity and equality, so people can use AI aids, AI assistant in teaching in China widely.
In governance, intelligent systems apply to dynamic, control, public services having significantly enhanced the efficiency in with the COVID pandemic.
And we also achieved progress in building intelligent public services system, the widespread adoption of AI in transportation, finance, and environmental being sectors have given rise to public service system, so if you know city there is a big lake in the center of the city, so the public transportation in this city five years ago this city (audio breaking up).
We have also enhanced the regulation and governance mechanism to ensure the safety use of AI and also China promoting green and intelligent synergy to leveraging AI to reduce the carbon footprint of social services and also production.
So in summary, China's endeavors in constructing an intelligent society are steadily transitioning from blueprint to reality. So the blueprint is a national plan. This transition not only mirrors technological advancement, but as underscores a pathway to refine social governance and augment wellbeing.
In the future, personally, I believe China should further emphasize constructing an intelligent society centered on people rather than technology. Efforts to empower individuals with technology, ensuring a dedicated balance between intelligent logical innovation and ethical operation.
Ultimately, the objective is to forge an inclusive, equitable, sustainable, and harmonious society for all by leveraging the potential much artificial intelligence. So that's my brief introduction of China's goals and actions in building an intelligent society. Thank you so much.
(Applause).
>> MODERATOR: Thank you, Professor, for your excellent presentation which provides a comprehensive overview of Chinas objective, plans, and actions in building an intelligent society.
Now, let us welcome Mr. Sam J who will deliver his speech on topic of international governance of AI and environment.
(Applause).
>> SAM J: Thank you very much. It's a great pleasure to be here with you all today. I've only got 10 minutes so I'm going to try to race through this quite promptly. First, what's the positive contribution AI can make to climate solutions?
Well, here are just a few. New material research in solar technologies, battery research, biodegrade balance difference to my colleague Philip Steer at oxford University has a leading intelligent earth project on this, and AI will be vital to achieve the new UN Climate COP Energy Efficiency Goals across all industries.
Well, what's the problem? Well, AI energy and water consumptions are high and growing. This has global impact as a contributor to greenhouse gas emissions. Currently, AI accounts for 1% to 2% of global energy use, but it's set to potentially increase significantly in the future. We need a new Japan worth of electricity every year because of AI but as because of air conditioning and electric vehicles, and the water use of data centers uses more water than four times that of Denmark every year. Other factors that could increase AI's energy use, there has been an emphasis to date on the energy cost of testing and training large language model, but in the future there will be greater emissions from inference, from multimodal research, and particularly important will be to track the energy use of semi autonomous AI agents. And generative AI will shift not just from shaping the Internet, but as relying on realtime IoT data of human behavior and natural processes involving greater data.
(no audio).
First, focus on the whole lifecycle of AI, from the mining of critical materials all the way through to the deployment of AI models. Second, standardize the way we measure AI emissions. This work has been taken forward by the ITU, ISO, IEC, IEEE and others, including at the recent New Delhi standards summit. Next, incentivize transparency from industry on their energy use and emissions. Incentivize efficiency in hardware design, make software more efficient, such as the work of the green software initiative, and data sobriety, more accurate well structured data reducing duplication, only use the data necessary for the task. And lastly powering new data centers, only using renewable energy and reuse components end of life.
But can we leave energy optimization of data centers and chip design to industry alone? Well, perhaps that might change. I say perhaps because industry has achieve remarkable progress. Data center energy consumption only increased by 6% between 2010 and 2020. Computer workload increased more. Kemp lar in 2012 to hopper in 2023 and Google achieved similar efficiency through TPU. But despite the efficiencies, AI energy use and emissions overall continue to rise. So AI is therefore subject to the jebins paradox named after an English economist who in 1865 observing efficiency of coal use led to consumption of coal across a number of industry, and AI is following a similar path.
So what are the prospects for a global interoperable approach on sustainable AI? Well, to have that, we need to navigate these geopolitical challenges. One, the greater U.S. China trade and national security competition. We've seen export controls, rare earth export bans, and so on.
Secondly, the new U.S. administration is moving away from green regulation. Thirdly, sovereign AI trends may make it harder to shift testing, training, and inference to other countries. And fourthly the New Bricks AI alliance announced by President Putin may lead to bifurcation approaches with the West. I want to end with what are the opportunities in 2025? Well, I think it's vital that we use fora that includes China as well as the West. So we have the UN universal tracks that come out of the two UN General Assembly resolutions, Responsible AI proposed by the U.S., co sponsored by China, and AI Capacity Building proposed by China and co sponsored by the U.S. Great initiatives and we can have further cooperation. Then the UN tracks are emerging from the Global Digital Compact and H lab on AI. Science convening, policy dialogue, standards, and capacity building all can be used to advance sustainable AI. UNESCO's ethical principles, ITU's AI for Good summit, and forums such as IGF. The Kingdom of Saudi Arabia, they'rette in ethical framework for AI and perhaps that could be a bridge involving China and the west. Malaysia chairing of ASEAN in 2025, they have an interest in Responsible AI to double ASEAN's digital economy. Singapore's had leadership on greener data centers in humid tropical climates in software and greater sustainability in AI verify and model frameworks. The EU and UK framework an reducing digital emissions, international agency's report next spring on AI for energy. The AI action summit, the UK international energy security summit, and Republic of keya hosting of APEC economic leaders. These are all mini let ral opportunities to further standardize approaches to measuring the energy cost of AI, but they can't replace global initiatives involving both China and the West.
The last initiative, is a really important one is COP30 in Brazil. I wonder, can we have a higher ambition coalition on this issue of middle and smaller powers moving towards COP30, possible national champions include Kenya, Singapore, UAE, Saudi Arabia, Kazakhstan, Brazil, and France. That's the end of my time. Thank you very much.
(Applause).
>> MODERATOR: Thank you. The energy and environment challenges in AI development are indeed global issues that require collaborative efforts from all countries around the world to address. Now let's welcome Professor Mi Jianig with online presentation from Beijing subforum, his speech is entitled Questions on generative artificial intelligence. Okay. Beijing, clearly?
>> Today I'm going to talk about questions about artificial intelligence.
(echo).
(reading slide).
Fundamentally it is transforming the knowledge production mode, mainly fundamentally changing of understanding the knowledge, truth, and cognition shaking the epistemological presuppositions of the subject object dichotomy of the supremacy of reason and have been established since the enlightenment.
Number two, does the emergence of generative large language models signify the end of anthrohyposen trism or does it mark the new starting point for reshaping human nature.
If intelligent is no longer the exclusive domain of humans, the creativity can also be simulated and surpassed by machines, then the subjective status of humans and the spirit of all things.
The next challenge is will the human machine collaboration from a new paradigm from the knowledge expiration. When AI is not only a tool for human cognition but as becomes a subject and even partner in knowledge production, the relationship between humans and machines will inevitably undergo a profound restructuring, that is number three.
Question number four, how will generative large language models form a new paradigm for knowledge exploration. The interaction between machines and human scientists to form amplified thinking will promote cross disciplinary and integrated researching, giving rise to disruptive innovations.
A new of medicine and physics has already shown that.
Then I will talk about the social restructuring, the transformation driven by generative AI, especially the decision making innovation, that is relevant with question number 5. Can generative large language models help humans break through the limitations of bounded rationality and achieve innovations in decision making? With the help of machineses' ability to extract insights from massive amounts of information, individuals will have the opportunity to transcend their own cognitive limitations and obtain more comprehensive and objective decision making bases.
Question number six, how will generative large language models reshape the structure of human society and relations of production?
And the decision making transformation is leading to the foundation of traditional social division of labor because the non differentiation of knowledge is acquisition triggered by large language model is based on the distinctions.
And then for question number 7, question number 8, how can we break down the barriers and construct a fluid foj graph without disciplinary boundary. So that is exactly what question number 7 is asking. Can generative large language models break down disciplinary models and construct a fluid knowledge graph without disciplinary boundaries?
The disciplinary classification system of industrial era is built on the basis of specialization and professionalization of knowledge production, and behind it lies the imprint of reductionism and mechanism.
And question number 8, how will large language models revolutionize social science research and create a new paradigm with greater explanatory power, predictive power, and guiding power?
With the help of large models, social sciences is expected to establish a integrated research paradigm of the data driven human machine collaboration and multiscale linkage, so that is very interruptive to the traditional way.
The two last questions about redefinition of intelligence values, the philosophical reflections triggered by generative AI. So for question number 9, how will the intelligence and creativity demonstrated by generative large language models redefine the human cognition? Because there is amazing creativity demonstrated by the large models based on the knowledge graph formed by training a massive Corpora, blurs the boundaries between the imitation and innovation, quantitative and qualitative change.
And question number 10, it is relevant to how the artificial intelligence is benchmarking towards the human intelligence. What are the connotations and extensions of aligning artificial intelligence with human values? What exactly it is aligning to. The question is not very clear yet. When the artificial intelligence system presents these difficult questions from unique perspectives, humans will be forced to re examine existing beliefs and achieve value upgrades through openness, neutral learning, and evolution.
So the human value system is also ever evolving, so that is the core on whether we should align with the plif ral part. Then in summary, generative AI intelligence is teaching is pushing the humanity towards a brand new era. In this era, the speed of knowledge iteration and updating will be greatly accelerated, and human machine collaboration will promote the flourishing of science, machine will help perfect social governance, and human machine interaction will enhance human insight. When the artificial intelligence becomes the norm, the singularity will no longer be out of reach.
In the new era, humanity will bid fairwell to a civilization centered on individual intelligence and usher in a new epoch characterized by collective intelligence. Everyone will have their own personalized AI assistant achieving self transcendence through human machine symbiosis. This will be a brand new era. This will be a new attitude, and that's why I put forward the 10 questions. Thank you. That's it for me.
(Applause).
>> I would like to express my gratitude to the organizers of the event. Just two quick points, while I will present the presentation, I have a large group that, that helps me on a number of these projects, and so this is not just my work, it's the work of my research group, so all the credit should go to them, and these are just my views, and they don't reflect any group that we collaborate with.
I guess we may not have slides. Is there a clicker?
Okay. So one of the things that I thought that I would focus on is to broaden the discussion around AI. As you see in this image, AI is just a small piece of this larger revolution that's underway right now on what we call cognitive computing systems. If you look at everything else around here, you will notice three things. Number one, AI is probably the most developed field among the among the collection here. So if you look at things like neural psychology, this is an emerging field. This is where we still have a lot of blue ocean. Whereas, AI has been around for decades. The reason why I'm showing you this image, is number one, it's very important to put AI in the larger context if you want to talk about building transformative societies. AI will have a role to play but it's not a major role that it will play. It will work with a large assemblage of other innovations and other developments. If you see everything on this image, you will notice one other thing. We are working at high speed when it comes to technical innovations in all of these areas, yet our governance and our frameworks to actually regulate and do innovation have large amounts of inertia. What I will do in the remaining few minutes is highlight a few key points that hopefully will stimulate some further reflection on your part. So if you can go to the next slide.
Perfect. So if you look at the other view of cognitive computing systems, you will see an image that looks like this. When we look at what really drives public value, we are trying to navigate these two issues of managing, governing actual behavior with cognitive computing systems and trying to understand what are individual's behavioral intentions. So if you look at behavioral intentions, you will see things like risk, you will see things like privacy. When you look at actual intentions, you will see things like trust, you will see things like social presence. These are the areas where, again, work is on the way, however a lot of this work is fairly disconnected from work going on that I showed you previously.
Okay. So if you want to go to the next slide. So in the interest of time, I will not go through each of these. I will just highlight one thing at the end. So if you look at transparency, and it's an issue that's plaguing a lot of governments, our research has found that transparency is a very nuanced concept. There is transparency in terms of how government achieves a given outcome, transparency in how we use technology, and then transparency in how government uses AI technologies. These three have different implications when it comes to explainable AI and our social license when is it comes to innovating.
Because I know we have two other speakers, I'll go to the next slide. So, one of the other areas, if we really want to build a truly global and AI or cognitive computing system driven society, we have to undertake fundamental work in terms of how inter dependent our information platforms are, how inter dependent our digital algorithms are. As recent examples have shown, if we have a single point of failure and it cascades around the ecosystem, we have actually increased the fragility of our societies we haven't increased it.
The other issue we have to do is if we want to cover how to get public value of this stuff, we have to begin tracking where is the money going. We have a long standing project where we have been looking the a where governments around the world have been allocating their resources when it comes to advancement of AI, and so if you go to the next slide, and with these three, I'll just highlight one point each. Right now a lot of the attention is on AI and larnlg language models. To me the technology is already out of the gate. It's very hard to regulate. It's very hard to govern when technology reaches a given scale. But we do have an opportunity when it comes to things like quantum computing. We need to get ahead of the curve, instead of like we've been doing with previous generations of technology.
The reason that I bring up the Indonesia example is we have a lot of countries around the world that have forgotten the classical hierarchy of needs. Many countries around the world are deploying large language modelings for the higher levels of Maslow's hierarchy of needs, when they haven't yet protected their databases and haven't yet prevented cyberattacks. It's a constant battle.
Lastly, one of the points that we haik in the report that's coming out is in order to truly reap the value of cognitive computing systems, we need to rethink how we design problems. So a very simple example is we are still trying to solve for health care in most countries, whereas the leading technology companies are solving for healthiness. They have completely flipped how they look at investment in healthcare and they are no more trying to solve healthcare, they are trying to build healthier individuals. But for governments to be able to do that, they need to restructure government defendants and restructure ecosystems. If we don't do that, I believe we will never truly realize the vague of the cognitive computing tools to make our societies more robust and innovative. Thank you.
(Applause).
>> MODERATOR: Thank you for the enlightening presentation. The analysis of cognitive computer systems has providedded us a new understanding for understanding AI and stronger human machine trust. Now, let us welcome Professor from public School of Management who will deliver online presentation from Beijing subforra and government and standardization development of the intelligent society. Please.
>> Good afternoon. At present the human society is moving towards a intelligent society. A new generation of
(two speakers).
Provide exploratory and cutting edge contributions. In order to use long term cross disciplinary and multi disciplinary and empirical methods, record, describe, and predict the ongoing and upcoming changes in society and leadership Professor Institute for social governance at the university and launch social intelligent experiment and explore in the past the social governance. In 2019, in collaboration with domestic and foreign experts and scholar, promote departments to be social governance experimental basis in 22 areas across the country. To largest knowledge, this is largest experiment on AI technology and governance on global scale. After 5 years of practice, the experimental govern nans has achieved many important results and continuously providing theories, technical, standards, norms for building intelligent society with human touch.
For example, the City of Ordos in northern China has community service platform by using small QR codes to cover service and operation operations using enthusiasm of public to participate in community affairs. The system and large scale mining AI based models have ininsured the safety and efficacy of co production promoting intelligent transformation of regional energy sector. For example, in the digital government, China Mobile has provided an intelligent customer service experience over 100,000 Q&A services to 30 million people with government affairs model, with 98.7% creating the most intelligent government assistant.
Based on the practice and vast land of China, we observe that the technical characteristics for AI are triggering a paradigm shift in the governance model. The humanoid nature, self learning, adapt anlt, human computer interaction and wide ranging social impact of generative AI technology has led to a triple change in AI governance. Number one, the transformation of governance object from material technology can subject to human technological subject, and static and stable technology to evolving technology and this requires us to pay close attention to the issues such as values, responsibility mechanism and copyright neck nism of the big model and adopt a flexible, open, and agile governance framework.
Secondly, the government's interface has shifted from only dealing with the relationship between technical elements to taking into account of the human machine interaction. This calls for strengthened governance of issues such as information, cognitive bias, emotional manipulation, and addiction.
Thirdly, the scope of governance has shifted from solely focusing on the process of technological innovation to emphasizing the macro system of technology society policy involving multiples a pcts, ethics, social risks, social impact, and this requires us to pay attention to the social applicability of technology and responsible development of AI.
With this shift in governance paradigms, we believe that standardization is the first move to address opportunities and challenges of the times and dealing with social culture and civilizations. Standardization is not only a political tool with technical attributes but strategic, leading, and social, and people oriented and clear trend of standardization, internationally has been shifting from technical standards to government standards, from standard refinement to standard prioritization. The main issues of standardization in AI have also expanded from traditional topics such as algorithms, data, network security, to comprehensive privacy, risk, management and social impact.
In recent years, China actively promoted standardization of intelligent social governance, the relevant departments studying and formulating governance for standardization of intelligent social governance to be the standard system for social governance.
In addition, the national standardization working group on the social application and evaluation of intelligent technology, SWG35, which is headed by the university as Secretariat and I served as secretary general and also conducted useful explorations and promoted formal establishment of standards including social impact, generative AI, social experiment, (?) will continue to promote the development of standards and social generative AI, at thing nolg, smart healthcare, smart justice, and smart grassroots governance. The future has arrived. We need to use standardized needs to promote we must adopt a pursued end, positive, and optimistic attitude to address the risk and challenges brought by intelligent technology. Let's join hands and promote the governance of intelligent society through the new paradigm of experimental governance and ensure that all regions can benefit from the intelligent society and build a people centered, humanistic intelligent society. Thank you.
(Applause).
>> MODERATOR: Thank you for the analysis of governance transformation and path for intelligent governance. Now, let us welcome the final speakers Mr. highly experienced computer scientist, he will show his insights on leveraging information and communication technology as a tool for sustainable development. Because of the flight delay, Mr. Ilel
>> Yes. Can you hear me? Can you hear me? Okay. Good morning. Good afternoon. Thank you all. It's a great pleasure to be at this session, and I just want to say that all the previous speakers that have spoken have basically addressed most of the issues that we are to address. I would like to start by saying that the basic principles of how we use information, communication, and technology, I'm talking mainly on artificial intelligence, it has to be human centered, and when it's human centered, we are also dealing with things that relate to trust and respect of human values. Maybe we put that at the center of anything to do with artificial intelligence, be it with various data models that we collect or governance structure and then we have a good basis of this.
And in talking about this, I would like us to go back to the final document of the governing for AI for humanity, which was released in September 2024, and by the UN AI advisory board, and remember that this UN AI advisory board was set up by the UN secretary general Antonio Guterrez in 2023 and the views do not reflect any organization or entity that they refer to.
And I want us to take I would like to read from a recommendation one, which I think is the basis of this session today. You know, and one of the key recommendations from that document, which is recommendation 1, was an international scientific partner on AI. It was recommended that this panel has to be diverse, has to be multidisciplinary in terms of experts in various fields. That's what this session has done. We've had issues with quantum technology and issues with using AI to mitigate climate change, which is a big issue in the world today.
But key things that we should look at, if we look very well at the recommendation 1. We have to have annual reports in terms of AI related capabilities, opportunities, and risks, and where there is uncertainty, and this has to remain the core trust of what we do in terms of does it serve and respect human values, is it really not encroaching of human rights.
We also have to look at producing quarterly and thematic research as UN bodies, that help AI, especially with achieving the SDGs. Taking from someone who comes from the Global South, we all know that in six years' time, we're going to be looking at the United Nations Sustainable Development Goals, and if we use AI in whatever we do, try to achieve no poverty, or health, or agriculture or climate change and by emphasizing on the UN Sustainable Development Goals 17, which deals with partnerships and cooperation. We'll be able to achieve all we talked about here today.
So, I would like, colleagues, for us to reflect on the human centric side of AI in what we do, especially with our young people, who are the ones going to be using this technology in everything that they do. And they are the biggest social changes. Our governments have to understand, our companies have to understand this. We have to start making sure that evidencal based research on the potential impacts of AI can make in the world we live in today. Thank you very much.
>> MODERATOR: Thanks for you for your wonderful speech. I find that although the speakers did not coordinate in advance, I note that there is topics that are highly complementary. And China's outlines and goals, and Professor explored the governance dimension in this context. And Professor Mi, experimental technological challenges host bid AI. And Kevin D, the government approach with AI cognitive systems. Mr. Sam highlighted the importance of global dominance in addressing energy and environmental problems in AI development, and Mr. Poncelet showcased the other side of AI in promoting sustainable development.
Due to time constraints, we are unable to proceed with further discussion and interactions. I would like to extend my heart felt thanks to all six speakers today for sharing their brilliant and thought provoking perspectives. Ladies and gentlemen, the future is already here, let us embrace it together. Thank you all, we are looking forward to seeing you the next year. Thank you.
(