The Future of Data-Driven Innovation in AI With Jerry Yang

In the early days of the internet, Jerry Yang co-founded Yahoo! And created an entirely new category of innovation. He then founded AME Cloud Ventures, a venture innovation firm that invests in seed-stage to later-stage tech companies. He joins Scale AI CEO Alexandr Wang in a fireside chat to discuss the future of innovation in AI. What is critical or foundational for AI companies to get right to drive innovation and be successful? What strategies or initiatives should organizations be thinking about to become more data-focused? What are the most important things that we all can do today to enable a positive future of AI?

Nika Carlson (00:23): Next up, we're excited to welcome Jerry Yang. Jerry Yang, co-founded Yahoo in 1995 and served as its CEO from 2007 to 2009. In 2012, he founded AME Cloud Ventures, a venture innovation that invests in seed Sage to later stage tech companies. Mr. Yang serves as a director on the boards of Workday Inc, Lenovo group and Alibaba group. Mr. Yang serves on the Stanford university's board of trustees. He's also a board member for the national committee of US-China relations, a member of the Brookings China advisory council, a member of the committee of 100, as well as a member of the council on foreign relations. Mr. Yang holds BS and Ms. Degrees in electrical engineering from Stanford university. Jerry is joined by Alex Wang, CEO and founder at Scale. Alex, take it away.

Alexandr Wang (01:25): Jerry, thank you so much for sitting down with us today. It's a pleasure to be able to talk to you and really excited to chat through everything from the early internet to AI.

Jerry Yang (01:35): Thanks for having me and, it's too bad we're not doing it quite physically yet, but hopefully soon.

Alexandr Wang (01:40): Yeah. 100%. So, I kind of want to start out with, obviously you were a very early innovator in the early days of the internet, in sort of the early internet boom of the nineties and created Yahoo. And your work really serves as a blueprint for, (a) what was possible with the internet, but also (b) kind of ended up influencing a lot of, obviously, the world events over the past few decades. So I guess the first question is like when you were working on the technology at the very beginning, what did you sort of view as the potential or how far did you think the technology would go and what's kinda surprised you about how the internet has evolved since then?

Jerry Yang (02:22): Well, you make me sound old, Alex. I guess I am dating myself by talking about sort of the early mid nineties and I think we all, even I, grew up on the internet in the old internet before worldwide web and the web as a protocol, HTTP became a protocol in I think 1989 and Tim Berners-Lee I think just sold NFT on his original code for a few million dollars. So definitely it's been three decades of the beginning of the internet, as we know today, which is, at that time completely decentralized, it had this promise of setting up a website once and then everybody with a browser can view it.

Jerry Yang (03:12): And that was really revolutionary back then, because before then it was all these other protocols like FTP and you have to learn all these commands to be able to get any information on the internet, but now you can just do point and click and I think, just like everybody else, my co-founder David Filo and I, saw some potential, like we'd be lying that we knew everything that was going to happen since then but it was obviously like super compelling to see the content that's emerging on the internet back then, to see the pace in which geographically it spread. And I think Yahoo was really an attempt by us to, at first sort of being categorizing websites, you're in the labeling business, we were in the labeling business literally building a hierarchy of ontology, a hierarchy of labeling and categorizing stuff.

Jerry Yang (04:16): And I think it was one of those things we never thought even in the beginning was going to be a business because there was no way to make money on the internet back then. So we did it as a passion and the labor of love and quite frankly, we were probably much more curious about what people are going to do with this content, is it news, is it sports, is it music, is it graphics and just about everything you can imagine started showing up.

Alexandr Wang (04:48): One thing I want to ask, like you've seen multiple waves of new technology and these new technologies ultimately fundamentally transform the way that we live and we work and now we're sort of at the Dawn or the precipice or whatnot of AI technology and have having that start to seep into everything that we do. Do you see common threads between AI and maybe the early is the internet or the early days of mobile? And do you think that AI can just as transformative as the internet in transforming consumer experiences as well as how enterprises sort of work?

Jerry Yang (05:28): Alex I think as somebody that grew up studying computer science and electrical engineering in the eighties and early nineties, that was always a dream. And I remember being in Japan in 1992 and people forget that that was sort of the information highway and Japan was going to take the lead on artificial intelligence, right? So this artificial intelligence has always been this holy grail for computer scientists and engineers and quite frankly it is always been... just felt like it was always one step too far for our current technology to absorb. And long behold in your era, in our current era, you have the confluence of the cloud, you have the massive number of data collectors out there.

Jerry Yang (06:18): It used to be all the data had to be human, but now our phones and our cameras and our sensors, our machines are creating as much data as people are and machine learning and newer networks became something that becomes an application that is very easily accessible. And I think it's transformative because I think all the promises of AI 20, 30, 40, 50 years ago are now in practice. And all of a sudden we're going, okay, what are the implications of it in a societal way? Are we really doing, for example, the real world work that you guys are doing on labeling objects and creating a tie between the real world and the digital world, that's unbelievably important and if you think about how the world will evolve going in the future.

Jerry Yang (07:17): I'm very host to Stanford university. I was in on the campus today and the research in RND that's being done there on AI is fascinating. And so it is probably one of the key technologies, if you look at the number of key technologies that we will all experience in the rest of our lifetimes, this is the key, right? If we do it right, it's going to benefit humanity in a massive way, and if we do it wrong, it could hurt the humanity in a massive way.

Alexandr Wang (07:47): Totally. Yeah. And what are kind of the, maybe the new and novel categories that you think AI now enables or alternative, what are sort of some of the things that you dreamed up maybe in the eighties and nineties when AI was a big thing, previously before the sort of AI winter that are now you've seen become possible. And how has that made you kind of believe in AI again?

Jerry Yang (08:15): Well, obviously, language is been the holy grail for AI. It's an application for understanding language, understanding words, understanding concepts, and to be able to generate and we're seeing massive developments there. It used to be, I remember doing an actual language processing, you're lucky to kind of get somethings right, and now have open source codes, and then you have private companies are developing massive vocabularies that are able to, not only understand language, but also generate language. And I think, we're still sort of on that uphill part of the curve. Vision is another killer app, I think of machine learning and AI and I think we're still early in that transformation where AI and Vision and Vision based learning is gradually replacing or extending our ability, our human eyes, vision, which is one of the most important functions.

Jerry Yang (09:23): And I think similarly, you're seeing there's a lot of work done on Vision, both in terms of throughput but also sophistication that can start really allow the world to see things, not just through text, but also through two to three dimensions. And then I think there we're seeing a lot of interesting, and probably the fundamentals are still improving. I think around infrastructure and ML tools. As an investor, we ended up looking at the infrastructure staff as something that I think needed a lot of capital, but where I think people are starting to connect the dots are the applications and everything from autonomous driving to the things that you see on the internet like if you like this, you might like this, recommendation engines and things like that.

Jerry Yang (10:16): But also, I would say in the last three or four years using AI machine learning in fields like drug discovery or life sciences are starting to really take human beings and take product and industry into places where we couldn't have been before without these tools. So it's really been exciting and it's still early days because you think about the problem of marrying our digital technology with the real world, whether it's navigation, whether it's logistics, these are all things that are still on the come and to me, it's one of the more, sort of transformative and more fundamental things that we have to continue to evolve the application side, but I think that the infrastructure and the basic tools and the foundations of AI still needs to be improved.

Alexandr Wang (11:09): Yeah, fully. Your VC fund, so AME Cloud Ventures invested in interesting company means from seed all the way to later stage, many of whom that are building infrastructure and value chains around data and that's sort of one of the investment strategies that you all have. Maybe tell us a little bit more, why do you have this focus on data in your investment strategy?

Jerry Yang (11:38): Well, I wish I'd met Alex you earlier, but I think it has to go back to Yahoo and I think people may or may not remember, but Yahoo was one of the originators of Hadoop and when we were trying to process the log files that we have, or just the amount of data that is to generate better servers to do ad, tech or consumer product improvements, I keep asking, why can't we just go buy a database to go do all this? And my team keeps coming back saying, there's no way, there's no nothing that can handle this volume. And so the idea of kind of big data was happening inside big companies like Yahoo at the time, I'm sure at Google and elsewhere. And so none of us had the commercial tools to go do it.

Jerry Yang (12:25): So we had to go develop our own. And when I left Yahoo in 2012, Alex, the idea that, oh my God, this is... whatever Yahoo is generating the rest of the world was going to be generating just as much, and then maybe even 10, a hundred and thousand times. So this idea of, it didn't take a genius to figure out that data was only going to get larger and more plentiful and that machine generated data is going to be the driver rather than people typing in spreadsheet or typing word documents. So I didn't know where it was going to go, but I said, look, I'm going to follow. If data is the new oil or data is the new life blood of industries, let's just follow it. Let's not have a prescriptive or a predetermined path.

Jerry Yang (13:13): And boy, it's take us all over the place. Like you said, it's helped us really get into places that we never thought be possible. Like I said earlier to life sciences, but even sort of food or ad tech, you think about things like impossible are all possible because of data and we have a great company called Zipline that they build their own drones that uses onboard AI to really navigate through places like Rwanda and West Africa. So I do think it's a very interesting premise, but like I said earlier, it took a combination of things like Cloud and Mobility and the ability to process massive amount of data in a very efficient way for this all to happen. And you look at AWS is a key driver to that. And so we're almost 10 years in from that sort of big data revolution, but I still think we're sort of the beginning of the curve on that.

Alexandr Wang (14:22): So when you kinda look back at a lot of the companies that you've invested in and you've sort of maybe pattern matched, what are the components of a data strategy that allows some of these companies to be successful or to ultimately build that data play into an advantage long term? What are some of the commonalities you see in successful data management, data strategies across the board?

Jerry Yang (14:44): It's a great question, Alex. And I think, obviously there's the sort of product ideation, the product market fit, and there we have been seeing companies that have great IP and great technical founders and very deep tech, but we also have seen things that are just scrappy and people kind of just really put things together that made something that work. And getting something to fit in the product market fit is harder than it seems and you've gone through some of this, which is, do you want a really big market and a product that serves a big market? Or do you want to find a smaller or a more honed in product that works for something that you can mine deeper?

Jerry Yang (15:31): And that trade off has always been a little bit of a trial and error in some ways, but also some companies kind of hit, they won and they run with it. And then ultimately the go to market, I think is the key, right? To your point that the data strategies only work if customers want it. And in our world, we've seen, products are not necessarily the best but if they can hit that go to market sweet spot, they grow like crazy and they can iterate and they can make the product better. We've also seen great products that didn't work because they missed a go to market window. And so, the good news about data as a business model data is that, every Fortune 500 now has a chief data officer, just about every Fortune 500 CEO I know I talk to is talking about how do I use AI and how do I use data strategies to change our business?

Jerry Yang (16:28): Everything from internal business intelligence, I can't imagine any company that isn't used and business intelligence to improve their business process, but also really, how do they put it in their products? How do they put it in their sort of day to day customer facing or internal processes to make it better. And I'll give you an example. I think when Yahoo was going through its growth phase, I would go to these big companies like General Motors or Walmart and it was very hard to do business with large companies as a startup.

Jerry Yang (17:08): Now, it seems like every large enterprise and certainly in America, is pretty accustomed in doing business with startups and I think that's a really important part for the business is to understand what's going on. We had a audio AI company that did a big deal for McDonald's to do their drive through, to take the order and McDonald ended up buying them, right? So this idea of big companies using these small companies as partners could sometimes lead to acquisitions or talent acquisitions that are really important for these businesses. So I think it's really been interesting to see how big businesses have adapted to new technology. If they can't grow internally, they'll figure out a way to partner and get it.

Alexandr Wang (18:00): Yeah. I think on these train of thought, one really interesting thing is, for most traditional enterprises, they were built and really grew in sort of a phase of business where data was much less of a focus. It was much more about operational strategies or [inaudible 00:18:18] strategies or acquisitions, whatever it may be. What strategies or initiatives, and you mentioned acquisitions, but what strategy or initiatives do you think traditional enterprise should be thinking about to become more data focused over time?

Jerry Yang (18:32): Well, I always say, there should be a very rigorous and top down meanings, sort of company wide mandate around understanding what data you have and then there... so that may involve everything from customer data to vendor data to supply chain data, to employee data on the board of Workday HR management system and just HR data in a competitive industry to hire people is gold, right? I mean, if you can anticipate, you can use AI to anticipate when Alex should get a raise in the next six months because he's been scoring high on all his reviews, I mean, you don't need a genius AI system to go figure out how some of these things ought work. So really a complete and honest audit of what you have and then where you need to be.

Jerry Yang (19:24): And this stuff now is more commonplace. You can probably get a best practice view of what sort of leading companies are doing with their data. And then once you have the gap, it's easy to [inaudible 00:19:35] try to address it. I think what people are always surprised at is they, they have a lot of data inside the company, it's just not consistent, it's not uniform in the way that you can act on it and that there isn't a plan or strategy to go and figure out how to use it over time longitudinally. So the good news, and I think the good news for scale out AI is that there isn't a lot of rocket science or evangelism that needs to be done. I mean, there's still some, but more and more people are getting converted.

Jerry Yang (20:10): And to your point or earlier, I think it is a little bit like the internet. I mean, you probably had to go around and talk about AI and labeling and the work that you do in the early days and people look at you weird, now I think it's more of, how does this help my business? How does this really improve my customers? How does this really retain customers? And I think that's a great stage to be at, because that means the industry's growing.

Alexandr Wang (20:34): Totally. Yeah. And in the same vein, we talked about the traditional enterprise, but like, what do you think is critical or foundational for sort of AI native companies or sort of new AI startups? What do you think is the responsibility, or the foundational work that AI companies need to do to get things right, and ultimately make sort of like an AI future possible?

Jerry Yang (20:58): Well, I think there is clearly... even in tech companies, there's always a debate of do we outsource some of this or do we buy? Do we own some of this ourselves or do it ourselves? And I think as we've seen in the more traditional internet sense, and certainly when the internet was growing, people felt like they had to do everything in house, but over time it was clear that, whether it's data centers or sort of middle layer software, or now you see industries pop up around security and operational software, and so it's my belief that I think as more and more AI tools and Maya processes are introduced into companies, people are going to say that this is our core competency, or is this something that we can find the right partner?

Jerry Yang (21:55): And it's important for companies that are in the tech, even though you're born in the AI era, that you understand the language that you understand data, the value of data, it doesn't mean that you want to do everything yourself and you have to decide what is core and what is not. And so that's one thing that we're seeing a lot of our companies going through now is once they reach certain scale, they're like, okay, well, do we keep investing doing this or do we do what we're really good at? And so even big companies like Facebook or TikTok or others that grew up in this world and certainly they have the resource to go do it, but they also realize they need to partner or have vendors that are really going to solve their problem for them.

Jerry Yang (22:44): And I think that's a healthy thing because I think the industry evolves when you can really see multiple use cases on multiple large customers that are willing to commit to this. And I think AI is a business around, not [inaudible 00:23:08], it's a business around scale, it's the more data you have, the more better you are, the better the results, the better the learning, the better the algorithms. So where people can really concentrate the learning and the data and run the fastest, that's where the advantage is. And I think companies that understand AI understand that, is where do we going to get our best advantages quickly?

Alexandr Wang (23:36): Totally. Yeah. It's a great point. Like, I think everything you're saying, which is like, hey, first, machine learning is not going to stay as this like one massive packaged up magical technology. It's going to be unbundled, just like how many technologies in the past have been and by building off of infrastructure providers or other technology providers, you're going to be able to move more quickly. And then the other piece is like, hey, it's really important that you focus on what is your strategy for continuing mass data, continuing to stay ahead, because that's ultimately it's going to dictate the performance of the AI.

Jerry Yang (24:14): Right. The quality of the data is so important, garbage in garbage out.

Alexandr Wang (24:18): Garbage in, garbage out, exactly. You're preaching the choir. One thing that it's impossible to talk about AI without kind of talking about China, especially looking at how AI is really seen massive developments in the US and in the Western world, but also in China as well. So, one thing I'm really curious to hear your thoughts on are, as we've seen sort of China launching pretty incredible results in AI and machine learning, recently there's a lab in China that announced sort of some of the largest language models ever be trained up. How do you think about, sort of the relative comparison of developments between AI in, first in the US versus China or the Western world versus China, generally speaking.

Jerry Yang (25:12): I think it's incredibly important and even within sort of the last five years, you went from kind of you seeing a lot of global collaboration around AI to now it's being divided geopolitically, and I totally understand it. I think one of the things that was a wake up call, I think to all of us in America, is, those of us who re realized how important AI is is that we need a real national strategy around AI and competitive strategy and France for example, have one, and I know Alex, you worked on it. A white paper that talked about how to maintain competitiveness in AI for the United States. And so once things became geopolitically divided, I would say that... earlier we talk about garbage in, garbage out, but here, I think, depending on the value system and the society goals that you have in America, for example, that's how we're going to train our AI.

Jerry Yang (26:12): I think AI is, whether it's deciding or not to give credit to a transaction or an insurance AI bot, or even language processing and censor and non censorship language models. These are all as the machine language models get more sophisticated, they are going to reflect the value that we want to create and if you think about how much debate and consensus building that we need to have in the United States already, can you imagine trying to figure that out with a country like China, which have very different goals and very different models. So I think you are seeing the separation of the two ecosystems around AI. I think China is intent on building their own capabilities around AI, everything from applications all the way down to the hardware.

Jerry Yang (27:06): And it's, obviously a competitive situation because I think everybody understands that if AI is a core part of your technology infrastructure going forward, there are a number of advantages in terms of the AI being able to benefit whether it's sort of business sector defense, government sectors. But also I think if you fall behind too far, then there's some significant disadvantages. I think that whether the two systems should be completely separated forever, that is not necessarily the best thing for either world, I think some common understanding, some agreement, some ability to govern how AI works is important because you can read the science fictions and see a world where AI became more powerful than all of us.

Jerry Yang (28:05): And obviously, at some point AI could be smart enough or being close enough in approximation to human intelligence if you believe in some of the things that we do. Now, I do think the us has tremendous advantages in our entrepreneurship, in our innovation, in our sort of decentralized way of creating value and I think China does it a little bit differently, they have a terrific entrepreneurial society as well, but I do think that the regular leaders play a much heavier hand. So we'll just have to see. I like our chances here in the US. I think we are more motivated when we think that we can create some good in society, but also have a very vibrant capital market, but I think China is going to be a real competitor.

Alexandr Wang (29:00): Yeah, no, totally. And a big piece of what you just said was something that I really deeply agree with, which is AI is not just sort of like in a vacuum, it's also the product of what are the values that we are using versus other countries that are using to build AI and those get embedded into the technology, what approaches technology we build on. For example, the United States, like there's a huge focus on privacy, preserving and AI or AI that minimizes the potential for biases and some of these other sort of more ethical issues and these are just going to be baked into the technology that we build in the future. What-

Jerry Yang (29:42): You mentioned that you're learning, right? What are you learning? Are you learning what is good and what is bad, right? And how do you define what is good and what is bad? And so I think the ethics of it is incredibly important and as practitioners of AI, businesses of AI, even though it may not be front and center, this idea of imparting some value system into the machine learning algorithms is something that I think we all need to... and of course, as they become core part of systems that run companies, the audit ability, the transparency of how did you decide that that person should get credit and that person didn't, and is there internal bias or inherent bias in the algorithms? I mean, we're biased as people because that's how we live. And so if we impart bias onto the machines, they will just amplify that in some ways. So, totally agree with you and I think that's something that businesses, academia, governments need to continue to talk about to get ahead of it.

Alexandr Wang (30:44): Totally. One related or a few related points. One is that, you obviously had seen and really watch this develop, kind of like the bifurcation of the US and the... or the Western and the Chinese tech ecosystems and the sort of consumer web. And that's something that sort of happened and in some sense it almost like is, continue to be like these two parallel worlds developing at the same time. What are some of like the interesting things that you've seen over time with that bifurcation and what do you think are these some things we can either expect, or we can learn from it as the AI ecosystems kind of start diverging?

Jerry Yang (31:27): Well, it's fascinating because I think when Yahoo was investing in Alibaba in the early two thousands, I think that the investment deal was in 2005, but the year or two before that we were learning just getting into the internet ecosystem in China. And, even back then, I would say, they would say it's a more or less a copy culture, right? Like, how do we emulate? Alibaba was the Amazon or eBay of China or the Xs and Y of China. And I think that happened throughout most of that first decade until 2010 or 2011. And then after that, you started seeing more indigenous innovation within the Chinese startup companies. They started making products, they started making changes. They started making products that were local and business models that were local.

Jerry Yang (32:21): And I think from then you see sort of that next phase a lot of internet companies in China that were, maybe started because they were X of Y in China, but then really emerge on their own, whether it's like Meituan or Tencent and Alibaba, which started totally differently, they ended up being huge conglomerates. Even Xiaomi I would say wanted to beat apple but ended up being a lot different. So I think that phase was important in China's industry development, because they were focused on their customers and focused on their core markets. I mean, I think going forward, it's going to be very interesting because I do think that there should be learning on both sides. I certainly think there's a lot to learn from the Chinese companies when it comes to short videos and sort of the use of AI in media and obviously TikTok is still a Chinese company, but they exemplify that here in the US.

Jerry Yang (33:29): And vice versa. I do think there are a lot of innovation in new sectors here in the US that I'm sure the Chinese are keeping their eye on. So, it is probably mutual learning that can happen and taking ideas from each other that are good. But the markets are evolving differently, right? I think the business models, I think in the US is still very advertising centric for some of these consumer facing companies, whereas in China, the commerce is just much more developed because they know of really had an ad business and now they're sort of doing this commerce enabled business model. And as business model differ, the different things that people develop to innovate is also different, but it's fascinating. And I think, probably for the next 20 years, we're all going to be talking about how one side of the Pacific versus the outside Pacific and how they do it, but I do think it's pretty vibrant on both sides.

Alexandr Wang (34:27): Yeah. I think you bring up a lot of... there are so many interesting points embedded in what you said, right? Like, for example the sort of diversions of the dominance of the ad based business models in the US versus the commerce based business models in China. That is implications for what kinds of algorithms got built traditionally, or what was the focus in AI and machine learning, and as you mentioned, like these sorts of differentiations create different opportunities in the tech ecosystems. Like one of my favorite examples or one I think the most interesting examples is Ant Financial in China versus a lot of how the FinTech ecosystem is being built here in the west. Ant Financial had this ability to be quite vertically integrated and therefore innovate extremely quickly on how to use AI in the process of finance and financial services and I think we have a different regulatory environment in the United States and so there's different requirements on how that [inaudible 00:35:28] the whole thing.

Jerry Yang (35:31): To your point about Ant, I think even when they enter in the market, there were payment but there were no real mutual funds and there were no real sort of consumer financial services. And so for them, it was easy to keep just going, just keep offering services and it was sort of a FinTech example that didn't exist anywhere else in the world. Now, I think, I feel like a lot of successful FinTechs in the west they have components of that, obviously different regulatory as you say, but it is interesting to see how the same principles are of growing your services on top of a customer base can evolve in different countries and how fast they can evolve because of AI and because of data.

Alexandr Wang (36:14): Yeah, totally. One common belief is like, hey, because China just has way more people and has way better ability to get lots of data, they have way better ability to label all that data, et cetera, they're just going to race ahead in AI, in most of the interesting fields of AI. What are you going to think about that? Do you think it's true? Do you think there's aspects where it's true? How do you think the US should adapt to that if it's true?

Jerry Yang (36:43): You're the expert as well, Alex, but here's what I think about it. I think that there's a couple of analogies people have used and there are truths to some of that. One is, US is really great going from zero to like a hundred miles an hour and that means all the Greenfield stuff, all the stuff that isn't necessarily kind of status quo mainstream people are contrarian thinkers, and people are sort of rebels in the way of thinking about it, it's like, okay, well, if you think I'm going to do this way, I'm going to try to do it this way, and I'm going to really try to do it this way. And so if you look at AI, it's still early enough, right? There's different ways of processing language models, there's different ways of achieving cognition beyond sort of the way that we've been doing it through current machine learning.

Jerry Yang (37:29): There are tons of different ways of marrying hardware and software that are people are exploring. There's huge push on a AI in the Edge and how you want to do that efficiently. So I think there's still a lot of greenfield stuff that I think America just seems to be better at because there is a culture of challenging incumbent and really being sort of the up starter. People root for the up starts, I mean, I remember people who want to see a Yahoo do well because they represented something new and different and better. And so I think that is true to some extent in China, but you can also see that there is obviously a government system that is really heavy handed when it wants to be and needs to be.

Jerry Yang (38:17): And that challenging the incumbent culture works only to certain extent there, it doesn't work all the way there. So it'll be very interesting to see what happens. Now there is obviously the argument that the more data you have, the more you're going to have advantage, and there's some truth to that, but I also believe there's critical mass set of data. That you need, and maybe there's sparse data. Maybe there's a world where you can do enough with sparsity of data that whatever that is. And by the way, I feel like as you get into the life sciences and everything else, there's plenty of data that we yet to uncover, that to say that we're out of data here in the US and they have more advantage, I think we're just looking backwards, we're not looking forward.

Alexandr Wang (39:04): Yeah, definitely. What other organization that you're involved in is... you're on the advisory council for Stanford's Institute for Human-Centered Artificial Intelligence, the HAI.

Jerry Yang (39:17): Yes.

Alexandr Wang (39:18): And of which Fei-Fei Li, who is also speaking at Transform X, she's one of the co-directors of. Why do you think the sort of human centered approach is important for the development of AI? And why are you so excited about that organization?

Jerry Yang (39:36): I think there was always this fear about technology becoming more than humans and whether there's more smart, more cognitive, you put that brain into a robot, they're going to do better things than we will. And I think the way Fei-Fei talked about it made sense to me, made sense to people who were more operating on the fear and pulled them into the side of advancing the technology on behalf of humanity, advancing the technology and understanding it so that humans can be better for it. So putting the human in the center around what our needs are, what our ethics are, how do we interact and how do we deal with each other, and then building the technology to extend those interactions to me made the most sense. She will be much more eloquent about it, but I think the work they do, deep on the core algorithm is still, I think, world class, but the real advantage of what she's doing at Stanford is this marrying of the technology with the political scientist, with the economist.

Jerry Yang (40:42): We have a great professor that is talking about the future of work, how AI can change businesses processes, or his analogy was that we're still using a 19th century factory layout instead of the horse and buggy pulling things, you put the steam engine in the middle of the room, but you still have the old factory layout. That's what's happening now with businesses. They still have the old business processes with the days before data running them. Now you just put in the AI engine thinking you can fix everything, but you could probably change a bunch of other stuff on the floor to make it even better. So as it relates to economics and businesses, as it relates to governments, Fei-Fei and the HAI group is talking to leaders in our governments and whether it's elected or our officials in the governments to really help them understand where AI's limits are, where our advantages are, where our disadvantages are.

Jerry Yang (41:36): So really integrating the notion of humanity into AI is something that I think is really powerful and quite limitless especially for an academic institution like Stanford, there's always a way to marry sort of the human benefits from AI. And if you look at it through that lens, it really identifies and prioritizes our work, so I'm really a big fan of hers.

Alexandr Wang (42:03): Yeah. Awesome. Well with that, that's thought to end it on. Thank you so much again for your time, Jerry. And, this is super interesting conversation, we read through a lot of ideas.

Jerry Yang (42:14): Thank you. I think you got more out of it than I think I gave you, but I'm really grateful that you had me on. Thanks.