Transcription
We'd like to invite Prof. Tanyang Kun on stage to deliver the citation of our distinguished lecturer, Mr. George. Seo, please.
Very good evening everyone. Mr. Giorgio has dedicated over three decades to public service and global business, navigating complex geopolitical landscapes and fostering international collaboration. Today he continues to be a prominent voice in academia and international relations. His ability to synthesize complex global trends makes him a vital contributor to shaping contemporary discourse on the challenges and the opportunities of the future.
Mr. served in a Singapore government for 23 years, holding pivotal ministerial positions, including information and the arts, health, trade and industry, and foreign affairs. His leadership in these diverse sectors has left a lasting impact on Singapore's national development and is standing on the global stage.
Following his distinguished career in public service, Mr. Yo transitioned to a private sector where he served as chairman and executive director of Kerry Logistic Network from 2012 to 2019. He's currently serving as a visiting scholar at the Lie Kuwanu School of Public Policy and a founding patron of the Asian Competitiveness Institution or Institute.
Beyond his professional roles, Mr. Yo's influence is felt across a vast network of global institutions. He serves on the boards and advisory council of numerous prestigious organizations including AIA group Pinto and the Burgan Institute on Governance. His expertise is also sought by the Holy Sea where he previously served on the Vatican Council for the Economy.
Mr. began his career with a distinguished military tenure, attaining the rank of brigadier general in the Singapore armed forces. He holds a bachelor of arts in engineering double first from Cambridge University and a master of business administration as a Baker scholar from Harvard University. We're really privileged to have you Mr. Yo. uh and the topic that you're going to talk about today is the challenge of artificial intelligence. Thank you, Mr. Yo.
[music]
It is now our honor to invite Mr. Yo up on stage deliver. Mr. video, please.
[music]
My doctors, dear friends, ladies and gentlemen, I thank SGH for the honor of addressing you this evening. My wife and I are very grateful to the doctors, nurses, and staff of SGH for looking after us and members of our family over decades. You know, SGH has a history almost as old as Singapore. You can't imagine Singapore without SGH. Without SGH, Singapore would be considerably diminished. And it is perhaps not surprising that SGH should be among the best hospitals in the world.
When I decided on AI as a topic for this evening's lecture, I did not reckon on [clears throat] the acceleration in this advance in in recent times. Just 11 days ago, Entropic introduced a new language model called cloud mythos. The release is restricted and only an initial version. An AI researcher was stress testing it. And to his alarm and surprise, he found that the model could virtually open every door and cupboard. He was testing the model for the ease with which he can steal things, spy on others, or sabotage systems. Entropic decided this is a bit too dangerous and held back on its release. And no doubt they were in conversation with the Pentagon and the defense establishment because Mythos is able to run through different critical software systems and find out what are his weaknesses. In one case, a bug not detected over 27 years was discovered.
Now, Pentagon, of course, you know, in the defense mindset, if there's a weakness, it's an opportunity. I cover my weakness, I exploit yours. So, the first step is to patch up their own critical infrastructure. So to do that there's a project called project glass wing which involves a number of the critical players. There's of course AWS web services. There's Apple. There is uh crowd strike and Cisco JP Morgan Chase, Google uh Linux Foundation because Linux is particularly vulnerable and then there's uh Microsoft, Nvidia, Palo Alto Networks and they gave limited access no doubt with the advice of the Pentagon to 40 other agencies responsible for critical infrastructure. Chat GPT immediately said that OpenAI immediately said that they're coming up with a model just as good. There is a race.
Now think about this a little little Singapore. Someone runs through our medical system, our power system, our telecoms, air traffic, find bugs. And this is what they call zero day vulnerability. Zero day meaning you have no idea that you have this vulnerability. So when it hits you, it's like a neural attack to which you may have no defense for a period of time by which time you may be in severe convulsion. This has created alarm all over the world. Imagine if you are in China or Russia or Europe, Singapore. Everyone is now scrambling. And yet mythos is only a milestone. The development of AI has become divergent. It growth is not exponential. What will it do to human society? One thing for sure, it is concentrating power and wealth in very few hands. You know the magnificent 7, these are the top AI companies. They already account for onethird of the New York Stock Exchange.
The latest issue of the economist if you have seen it, it showed the new gods. There's of course Dario Amod from Entropic. There's Sam Alman from OpenAI. Then Mark Zuckerberg from Meta. Uh Damis Hannabis from Google and Elon Musk. On the cover they dressed as Greek gods. And the question in the economist is do we entrust our future to them? Now is a complex figure. Ish is a colorful figure motivated by by money of course but more than money by power by fame by vanity and maybe by moral purposes. Maybe You know when Palanteer was using cloud to conduct surveillance and to put into weapons systems. Dario Amod said no we won't agree to this. You must put a human being in the loop and we do not agree to the surveillance of American citizens which means that the rest of us are fair game. Peter Hex says the Pentagon was very angry with him. He says this has created a supply chain vulnerability and wanted to cut cut him out. No, there is a whiff of hubris. They are now able to track everybody anytime, any place in any situation.
It began with the Ukraine war and Silicon Valley was helping Zalanski because many of them had their roots in the Jewish peer of emancipation. who the Russian generals were and they were picked up one at a time until the Russians were able to mount the response. And drone warfare is not just simple drones. There's a lot of AI and EW electronic warfare in drone warfare. Then during after October 7 when Israel was hunting the leaders of Hamas, they were assisted by Palanteer and just by monitoring electronic traffic. They knew who was where with a high probability any time. And the models are weighted to go after that person and to ignore collateral damage. It is quite scary. It was repeated with his bullah first in the pedra attack which was primitive and more recently when Israel immediately after the ceasefire managed to catch a group of his leaders on a zoom conversation within 10 minutes they knew where everyone was and hit them all in the city of Beirut causing hundreds of deaths.
Last June when the US and Israel attacked Iran, suddenly drones assembled in Iran, flew up, came outside the window, looked at you and zapped you. The Americans were so confident after their success in Venezuela, they thought they could repeat it in Iran a second time. And those of us who followed the operation in in Venezuela, you have seen helicopters flying very low and unmolested. Even though the Venezuelans had meant many portable air defense systems, selectively they were either destroyed or paralyzed. Secret weapons were used to kill the Cuban guards. And Maduro and his wife were plucked out and sent to the US. And today, despite protestations to the contrary, Venezuela has to do everything America wants it to do. Now, I'm not sure how long it will continue, but sooner or later there'll be a a reaction. They thought they could repeat it in Iran. So knowing where everybody was suddenly in a simultaneous simultaneous attack involving thousands of targets including Herman the supreme leader flipped the leadership and the chief regime change. But unfortunately for them, the Iranians learning from last June had already created a counter strategy where they decentralize their command that even though you hit the brain, there is a mosaic distribution which means that when you're negotiating with them, it's equally difficult because there's no one brain to negotiate with. And the supreme leader who was in his 80s deciding that rather than cowering in some deep bunker, he had a whole meetings in the open and was prepared to die, become a shahit, a m for some strange reason. The Americans and Israelis did not understand the Shi mentality of martyrdom.
Now, if you visit Shi shrines, they continue to moan for Ali and for Hussein who were killed a,300 400 years ago. In 2010, my wife and I visited Damascus. Before the unraveling, we went to the great mosque. And inside the mosque, there was a Shia shrine honoring the head of Hussein, the grandson of Prophet Muhammad, the son of Ali. He was killed in Kabala. The devotees there divided into men and women were still weeping. But when they saw that my wife was there, the man quickly parted like the Red Sea to allow my wife to accompany me to go right up to the railing, touch it presumably to gain married because they wanted us to get that blessing. So whether it is Nasraa or Solmani or Ham or Larijani or Sinoa the Hamas leader they've all become saints in the Shia pantheon Sunni but the others Shi those of us who are Christians know [clears throat] of the 12 apostles of Jesus Christ apart from John who had to look after the mother. All the 11 were muted and in the end it was the blood of the marts which one for Christianity the Roman Empire. So by killing harmony they made the problem worse.
So in this dynamic you see the difference between artificial intelligence and intelligence. You know when they were negotiating Islamabad, the two leaders, the speaker of the house and Arachi, the foreign minister of Iran, they were afraid for their lives even as they were negotiating because previous negotiators were killed during negotiations and even as negotiations had not finished. Washington Post in an opinion piece said if they will not negotiate to our liking let's kill them let's only leave a life alive those who are prepared to negotiate with us so I was following Mandi's Dr. Marand this podcast. They were so worried. They left Islamabad surreptitiously. There were other decoy aeroplanes. They quickly landed in Mashad and from Mashad they returned back to Tehran by road and rail. You know this hubris is scary.
We would have remembered what Edward Snowden revealed. He was involved in it as a geek developing the prism system where everything whether it is your phone, your car, your computer will locate you. He was so troubled he decided to blow the whistle. He went to Hong Kong and now he's in Russia. There was a period when the top people in AI in America were so confident, you know, for a while they were worried that China will catch up. Then when they made the advance in AI, they thought no, China can't catch up now because in the AI stack from infrastructure to hardware, software, applications, they dominated. They only worried about one thing if there's a nuclear exchange. So persuaded by Eric Schmidt, Kissinger went to CCing Ping in Ping to get the Chinese to agree to putting a human in the loop in the event of a nuclear exchange. But for AI generally, they thought by pushing the American stack into every country, prism will be brought to an allgether different level. Now of course life will continue normally but at critical moments when it is in American interest they will have an information advantage by selective release of information to tilt any playing field and occasionally to blackmail.
But luckily for all of us, China came along and suddenly we had Deep Seek which could get by with far fewer chips and somehow more efficient in creating an AI system comparable to what the US had. Sam Alman immediately said, "No, no, no. They were distilling from us and really benefiting from what we have done for their own systems." I'm not so sure. But China made a critical strategic move which was to make large language models open source. I do not know who started it whether it was deepseek first but in any case it became policy. So the other Chinese systems whether it's the Chenguan or Kim or Miniax they're all open source and they will catch up with MTOS and after MTOS came up they interviewed Jeffson Huang from Nvidia because they're trying to tell Jeffson Huang don't give China even the lower chips anything which helps them is not in our interest. So JSON said the Jensen said you know for the Chinese to do what MTOS does they have more than enough compute capability today. So they will catch up and he reminded them that in terms of AI publications, AI experts, there are many more in China than they are in the US. And even in the US, many of them are ethnic Chinese. And I was listening to this interview by Dikshu Patel or something. The no by somebody else. The the innuendo is that you are Chinese. So therefore you're talking about US and China working together. But you should not because the US must be ahead. And in the chips we have a choke hole. They're not realistic. China has many choke holes on the US. I mean, imagine if China were to somehow blockade Taiwan. If there's war, there'd be no Nvidia chips. But China has alternatives. China's ability to choke the supply of Nvidia chips is more potent than the homo straits if it comes to it because they have absolutely no alternative and there are many critical minerals materials which China can deny to the US. So Jensen Huang was being realistic.
In the end, is it really a fight to the finish between the US and China? And if there's a fight, can you win? So anyway, Trump says he's looking forward. He say Ping will welcome him. I think they will meet. They will have a certain resolution of the conflicts because right now the US cannot afford a conflict with China and China having played certain cards like rare earths have signal to the US that it will not buckle but at the same time it will not escalate and I hope that India will also become an AI center. Not two months ago they had a big event in India about making India AI power. It was not very well organized. So people laugh at it. But India has brain power and sooner or later you will get the chips and sooner or later you will learn from others and you will develop the capabilities. So the diffusion of AI which China has started I think is good for the world.
AI is not valuesfree and those who control it can weigh in such a way as to influence all of us sometimes in ways that we are not even conscious of. One day I was in Tinua and I went I returned home to Singapore by Tasing Airport which is the new airport. Took an hour. So Tinua undergraduate accompanied me. We were chatting. He said he has to use a VPN to access Chat GPT. I said why do you need JPT? You've got deepse. He says you want to find out anything about the Chinese government, you got to get out. But it's the same with the US. If you want to find out anything about the deep state in the US, they won't tell you. And for the Muslims, if your AI will print you pictures of Muhammad, I think they'd be very upset. There's no avoiding but the control of weights and instructions. the way AI is designed and used. But there's no doubt that we're on the threshold of a new epoch in human civilization. Everything which exists today which make possible the world that we live in today whether it's agriculture or industrialization or the internet everything will be transformed by AI you will create upheavalss the concentration of power is not a good thing. Human relationships will change.
You take say the medical your profession. I was chatting with a doctor who is now in the Macau University of Sense and Technology. He had just come back from Harvard. He spent time in Johns Hopkins. He told his mother who is in the 90s. He said, "Mom, if you live five more years, I'll give you 50 more." 50 more. See, Sting Ping was overheard talking to Putin about living till 150 years old. And everywhere people are talking about longevity because people are seeing what technology is able to do. Another doctor in Macau told me he's reverse he's genetically modifying a pig. A pig which has organs very similar to us so that the pig can be a source of blood to everybody can generate the eyelet cells. He says don't worry about type two diabetes. We all worry about tsunami in Asia. He says no this should be overcome. In every field of medicine, people are talking about a new frontier, which means many of us, particularly the younger ones, may become jobless. In a society when unemployment grows and people see AI displacing their work and smashing their rice bowls and a few people becoming wealthier and wealthier. [clears throat] the social stresses will inevitably build up. Which means like in the 19th century when industrialization created the revolt of socialism and communism, there will be in every society a reaction against the speed at which AI is disrupting normal society and our experiences, our collective wisdom are far behind. the advances in technology, our laws are far behind.
I remember talking to Wongen, our eye doctor, who's now in Tensing about two years ago. He said, you know, for diagnosis of eye conditions, AI is superior to any eye doctor. Question is, if you don't use AI, are you culpable? I I think so. I mean, if if you are a doctor and there's X-ray machine and they don't use X-ray machine, I think you are culpable. If you use AI without human being in the loop, are you culpable? Well, there's no easy question and in every field of medicine, this question will have to be asked. And when they litigated, judges following a common law tradition will have to apply new wisdom because it's a new frontier. So in every field of human society, there will be a revolution. These are dramatic times. Some society will be able to adapt, look after the weak, soften social stresses, take in the technology, improve productivity, search ahead. Others will be quaring among themselves. There'll be a lite reaction in many places, machine smashing of a new kind and they'll be trapped in the past. So both between countries and within countries we are going to see new tensions.
If you look at previous major shifts in technological development, they mark periods of history. Once you were agriculture and enormous surplus, you could organize big armies. You can create empires. Once you had industrialization and the assembly line, you could kill and slaughter on an industrial scale. Today with AI, when it's applied to warfare, the horrors are unimaginable. I do not know whether you've read the threebody problem? is part of a trilogy. The second volume by Lu Hersh was on the dark forest and it talked about a virus genetically engineered to kill a particular individual. Can it be done? Obviously it can be done. I mean you look at the co virus. I think many of you would have read Jeffrey Sax. He checked the Lancet Commission on COVID. You line up all the corona viruses and you can see how they evolve whether through bats or other transmissions. Suddenly you come to the COVID 19 and in the flu cleavage so many mutations they could not have happened naturally unless you have missed out many corona viruses in the world. So he said it was engineered increase in function for what purpose? The reason given is of course to improve health care to ward up future epidemics. But inevitably every new technology that arises will be pressed by defense people to gain an advantage in warfare. I was in defense for many years. I was head of air force planning. I was director of joint operation and planning. So I I had to think like that you know. Then what? In the book, they had no response. So they froze him for a few centuries until the therapies catch up to keep him alive. But that's science fiction. But that reality may soon be upon us.
How do we respond? Philosophically AI is not a god. Yes, it can do things which none of us can do but it cannot go beyond the sum total of human knowledge and what you can deduce or induce from it. New knowledge requires contact with reality. So you can create an AI model of the moon. But until you go to the moon and dig and build and somersault there, there are things on the moon which however advanced your AI model is cannot capture. You need contact with reality. A lot of the contact will be done by machines and AIdriven but because it is frontier the human being will be in the loop. In any case, as a theoretical argument, what is unknown in the universe is infinitely greater than what is known. And AI cannot go beyond what is known. But who pushes that frontier? Just AI or AI with human beings? So the question that we've all got to ask ourselves is what is the nature of human beings? Why are we spiritual? Why do we appreciate the arts? Why does music affect our mood and our productivity? You can get AI to produce very good music. maybe better than most musicians. You can produce the best music, but when your own child plays a piece and it's quite good, something magical takes place because that child has become more human.
They say now and it's increasingly true that you don't have to learn how to program. When I was in university I learned for trend some of you may learn cobalt all that is forget it there's no need to I still keep my old engineering slide rule I look at it I said yeah you know it was interesting but it has no place in today's curriculum. So learning computer languages may long no no longer be required but what is the highest computer language it is human languages that is the highest computer language and human language is not just instructions human language is poetry is literature it's Shakespeare and it's expressed differently in different languages is so when we talk about the education system how to impart knowledge to the child yes we should not burden that child with formulas and so on because all this he can access readily but poetry language sensitivity it's one thing to say give an instruction to your open the window, it opens the window. But if you don't form complete sentences, it doesn't understand you. So if you speak perfectly, it knows what to do. If you speak in poetry, you'll be confused. But one day, if you want it to do subtle things, you may have to speak to it in poetry. And its responses may be equally subtle. and sometimes unexpected and surprising. Which means the education must be what makes people more human. And it comes to the fundamental question. Is AI there to diminish us as human beings or is AI there to make us more human? And in this reflection, our spirituality will in the end come back.
This has implications on education, on the way human society is organized. With methos, it's quite clear that centralized systems, hierarchical systems are vulnerable. Nature has experimented with all possible combinations and complex systems inevitably become more and more biological in the pattern. And as AI makes human society more complex because suddenly we are paralyzed, we kill each other, many will die and then we create a response like the Iranian response. We create mosaic system, decentralized intelligence, an attack is viral and then the immune system recognizes it reality and counters it and it's an endless game. This is a biological world. Now human society will eventually become like that. No longer one based upon agriculture or on industry but one based on distributed intelligence. A little like biomes. You know there are big biomes which create organic entities. There are microbiomes which move around adding or subtracting those big entities and all in flux influencing each other. sometimes attack, sometimes attacking. But this is a time for reflection. It's a time to reexamine inside us. What does it mean to be human? If somebody is affected, do we discard that person? If we do, then that person is no more than just a bit or a bite. as we will look after the old or the because he or she is human. So this must always be woven into our approach towards AI. How to make the human being more human which means to change him from within. It makes a big difference whether you know AI can produce you a beautiful speech or beautiful PowerPoint but is it from you and we can all sense it whether you are just reading or paring because if it's from you the emotions will show there'll be mistakes and in that phenomenon we see a reflection of ourself self and that reflection in each other is what creates society and what makes us human. Thank you.