Name: Fortune Global Forum 2023: Beyond The Horizon
Uploaded: 2023-11-29
Duration: 17 min 40 s
Description: Peng Xiao, Chief Executive Officer, G42 In conversation with: Matt Heimer, FORTUNE

Peng Xiao, Chief Executive Officer, G42  In conversation with: Matt Heimer, FORTUNE

Fortune Global Forum 2023: Beyond The Horizon

Conferences

Peng Xiao, Chief Executive Officer, G42 In conversation with: Matt Heimer, FORTUNE

So there's a lot of buzz here in Abu Dhabi and around the world right now about what's being called the world's largest artificial intelligence supercomputer. The system was launched earlier this year by the Abu Dhabi Born tech company G 42. And it is said to offer a new approach to A I computing that's going to significantly reduce the time it takes to train an A I model. Uh In addition to the supercomputer G 42 has also unveiled J which is an open source Arabic language learning model considered to be the most advanced in the world. At the same time, events this week have shown how very geopolitically sensitive this kind of advanced technological work can be uh to discuss all this. We're very pleased and fortunate to have G 40 two's CEO Pen Chow joining us today to talk about his company's innovations, uh its mission to be a global leader in artificial intelligence and the challenges and responsibilities that come with that mission. Uh So please join me in welcoming Pang to the Fortune Global Forum. Please have a seat. So paying for people in the audience who aren't already familiar. Uh tell us more about G 42 and uh what it does and uh what you're, what you're hoping to accomplish. I got the back right here and thank you for bringing me up. Right. Yeah. Uh Anyway, so, yeah, tell us more about G 42 and what your goals are and what you're trying to accomplish. Thank you, Matt, first and foremost for this opportunity to be here with you. My pleasure. Thank you. As we discussed uh backstage, there are so many great and important things happening in the field of A I. Uh I I cherish this opportunity to discuss with you. G 42 is a company that started back in 2007, 2018 time frame as our friend Sam Oman who was here recently said Abu Dhabi was on A I before A I was cool. So uh that was part of us uh transpo this brand new technology becoming a global game changer. So we embraced A I as a core technology development objective of our company and build our business around it since then. And the company has developed into uh from the original research mission on on foundational A I work into many different disciplines across different industries, from health care to the space program. And here we are today. What are some of the specific challenges within those categories that G 42 is uh applying its uh its technology to? There are many specific industry specific challenges. But uh there, there are some fundamental challenge to A I uh development in general. Uh Specifically, I would say three things. Number one, you need a great talent, you need a great scientist and that's really the brain of creating A I. And second, you need a lot of food to fit into the growth of A I. And the food is, of course, the data and the third element is you need also tremendous of muscle to process all this science and data that is compute. So we've been tackling all three areas of challenge for many years now, attracting talent here and working talent with talent globally as partners and working on safeguarding and also utilizing large data sets. And finally working on supercomputers, you mentioned earlier in opening uh talk to bring more muscle to this uh to this effort. Yeah, thanks for bringing up the supercomputer. Um I think generally when people talk about A I, they talk work, uh the work being done, the models being trained and applied uh in the cloud in a sort of distributed way, uh The super computer seems like an opposite sort of more concentrated, physically concentrated model. Uh What are the reasons for going in that direction? What advantages might have confer to what G 42 is doing? Great question. Thank you for this question. Uh This is the reason why, by the way, we, we, we undertook our own super computer initiative with a partner therapist in California. You're right. Traditional hyper scholars use cloud in structure to help you to build your models. Uh It's a good way of developing model because they have so much compute. But it is also challenging the sense that we call system engineering effort because it is spread across distributed. You have to build your model in a smart way to take advantage of this vast array of resources in comparison if I can show you here, this is time for show and tell and it's good. I love this. It's very tangible. This is a single chip size of a small pizza that our partner service in California developed on this one chip, we have 2.6 trillion transistors. You can take a look. And so the advantage of this of having it all in one place is the speed up the training time, 100%. So literally 100%. Well, not literally actually, it can be more than 100% much more. So basically, what happens here is that in a traditional environment of many smaller chips across different systems, there's a huge cost of moving data on of the chip. That is where things get to slow down, that's a bottleneck. But if you can load all the data, you want to train on a single chip without having to an off ramp all the time, you save tremendous of of time. And that's why we believe in this supercomputer training model. So the one we launched, you refer to uh just now called Condo Galaxy, Condor Galaxy. Yes, phase one of Conor GLA was launched in California back in, I think July, it was already four X flops of capacity. So four X flops is in layman's terms. Uh one single A flop is 10 to the power of 18 computational cycles per second. It was very fast. It's very, it's a very large number. So we had four of those and we just recently announced in the uh in a high performance computing show in the in the in the us that we're doubling that now to eight X. So the whole idea here is to speed up the training of large models with potentially trillions of parameters, not just billions. And this is why we pursue this architecture. So Cerebri is an important partner, they're making your hardware. Uh Your company has forged a lot of relationships globally. Um What role do those partnerships in those collaborations can play in your strategy as you're assembling data as you're figuring out problems to attack? Like what are some of the key, some of the key partners that you're working with? Yeah, server for sure is one of them on the hardware side. By the way, they also co develop models with us too. Uh I heard you mention that this model, the largest and most comprehensive, most accurate Arabic model in the world was actually building in California between our researchers in Abu Dhabi and researchers in California with the um in a partner, we look for two things, number one superior capabilities than what we have. They bring to the table. And second, they have to share our value system, we have to have the common belief. So for each one of the problems we have, doesn't matter with service or with our friends at open A I or to global giants like Mac Microsoft, we find both. We believe there is an urgency in developing not just A I generative A I all the way to, in fact, general A iag I, we share that mission, we share that urgency and second, they bring to the table capabilities we actually don't have here. And even if you want to develop, it will take too long. So for example, with Microsoft, it's skill, they have skill A I is all about skill. You cannot develop A I without that large scalable platform. So we choose Microsoft as a key partner in that area. Open A I, of course, is a key research and development partner. We actually recently announced our partnership with them. We were joining the road map together from our research ecosystem here in Abu Dhabi, including our M BC University of Artificial Intelligence, together with Sam's team to ensure we don't duplicate our conflict. We go down the same path together, hopefully moving faster. And again, in the case of therapists without the hardware, we're gonna go nowhere. Sure, sure. That's the value that brings to the table, your partnerships uh gonna kind of address the elephant in the room. Uh Now your partnerships this week have attracted some some negative publicity. Uh Earlier this week, the New York Times reported that us intelligence agencies including the CAA had been raising concerns about G 40 two's ties to Chinese companies. Uh summarizing the report. The story essentially said us officials fear the G 42 could be a conduit by which advanced American technology is siphoned to Chinese companies. Um At the same time, the same article said, uh and again, I'm quoting uh companies in almost every sector around the globe have ties to China. So you do have some Chinese relationships and Chinese partnerships, they're attracting some concern. Uh There's also this group geopolitical tension between two major nations that that many people in this room, many business leaders have to navigate. How are you navigating it? How are you uh how are you dealing with this new attention? How are you, you know, confronting the the reality of, of this difficult environment? Thank you for the question. Thank you for the answer. Um First let me acknowledge my own ignorance because I didn't actually finish reading that article. OK. Someone sent it to me and something more important happened. I just didn't finish reading it but I did get a gist of what New York Times was trying to say. Uh as you mentioned already, Matt, this is a very strategic global topic A I and it has far reaching consequences uh for our society here in Abu Dhabi e to geopolitical situations around us all the way to the entire humanity. So it's correct that we all need to pay attention to the proper development of A I. And ultimately what we use A I for to that end, we have been a leading practitioner in this area. Uh As I mentioned, you saw our list of top partners, we are working with the safety of A I has been always a top priority for us. We spoke earlier backstage, the importance of safeguarding your data, the importance of proper data governance. I think that domain has been more or less dressed addressed pretty well from the previous work we did in big data analytics from the days when Google launched its first search engine. So when you say it's been addressed pretty well, are you saying that there are frameworks in place? Yes, that require you to the classification, data privacy, data residency. All this I think the age of hyper scholars, they are I believe fairly well addressed. And when you say well address, do you mean there's a wall built enough that is firm enough to keep data from being in the wrong hands. They are good policies and procedures in place. It's up to the practitioners to implement them in the right way if they are implemented like in our case properly, there's very little risk of data privacy concerns. However, I think you're referring to me probably going beyond data, the safety and security of intelligence. When data is actually put into a model, when the model has been developed, when the model begin to generate intelligence, this is where we are in a brand new territory. And I fully agree because it's brand new because it's so nascent, we have to as practitioners work together hand in hand to safeguard. Number one, the proper generation of intelligence, how do we know it's the right form of intelligence? How do we know it's really safe? Give us the right answers. And number two, how the intelligence becoming more and more powerful will not fall into wrong hands used for wrong purposes. Does this mean it's, you're not only responsible for safeguarding the data in terms of where it is held? But do you also have some responsibility uh over how your partners uh or I should say maybe over how your customers use it? And can you control that? Absolutely. I think uh I'm not looking absolutely that that's the right right uh question we need to address it with our partners met for sure. We have mutual agreement on how data is handled. For example, when we ship our Arabic data all the way to California to be trained on condo supercomputer, we make sure there's no leakage of sensitive information from our data set leaked in the US. The same thing if the US counterparts ever share any data with us, we are accountable to ensuring that that data stays safe and sound here and there's auditing procedures, this remote monitoring procedures in place for that. Now, the challenge at the moment when data or models get into your clients, your customer hands where you don't have that type of control anymore. Of course, your customers will sign certain agreements but ultimately how they use it, you as a vendor cannot be fully accountable for it. This is why we need a overall ecosystem framework from government all the way to consumers. And for that, you need, I assume collaboration, you need a lot of collaboration and transparency that's quite difficult. There are two arenas in which G 42 works if I understand correctly where there's particular concern around how data is used. Uh One is in uh medical research and essentially uh drug development where individually possible genetic data could get out into the ecosystem. Um The other is surveillance. There's uh I believe your company pres site A I which you know, does security and surveillance technology that police forces use. How do you make sure the same ethical standards uh apply in those in those even more sensitive realms and how uh you know what additional burdens come with that? And uh are you fairly confident again that the, the right safeguards are in place? Yeah, you just, I think two questions here. I'll try to answer them both. In the case of presci one of our companies in the T 40 portfolio focus on big data applications. We actually provide to our clients, the big data platform and the capabilities on which our client can develop their own applications. If our clients build applications commercial in nature is non sensitive, we build it for them following their guidance and their requirements. But in cases where the client actually are building sensitive applications continue extremely sensitive data sets and use cases, we train them how to use our tools to build their own solutions, which is in some cases, uh what preside is doing with some of our government clients here in the case of genomic data, you know, there is a Supreme Council of genomic Data Management here in the country. So there are multiple layers from regulation down to it policy to safeguard genomics and other other medical data sets in the country. In fact, I can tell you right now, I personally have never been able to get my hands on any of the data sets in the WTS database we have in the country and we don't need to, that's not our job to do so the data belong to the individuals, our job ultimately help our nation, help our partners to bring inside all of that data. Then to go back and provide hopefully precision medicine support to the population to do that. You actually don't have to disclose or even in many cases access, some of the sensitive you referred to. If I heard you correctly, he helping our nation. You're a, you're a national now a citizen of the UAE. Yes, I am. Uh What drove that decision? Well, the work I was doing in the UE is very important to the national agenda. The national agenda being to decarbonize, to diversify the Absolutely. Absolutely. When I came here almost nine years ago, I was answering a call to action from this nation that is they want to go from an extractive economy of relying on oil and going to a creative knowledge economy. That's why I came to help to build this big data platform, which eventually became this A I group of ours. So it became clear to me this cultural action became my calling, this became home. And because I'm doing such important cutting edge work in this country, I was very happy to join citizenship of, of UE. And the regulation here is such that you can only have one citizenship. I see. So here I am. Yes. Well, you answered their call. We're really grateful that you answered ours. Thank you for coming and talking to the global forum audience. Of course, good luck with the work you're doing. Thank you so much. Thank you.