Building An AI-Powered Society with AI Fund's Andrew Ng | Greymatter

Foreign Hi everyone, my name is Sarah Goa and welcome to Gray Matter. In this episode we feature Andrew Ng, one of the foremost thinkers on artificial intelligence. He is the founder and general partner of the AI Fund, co founder of Coursera, former founder and lead at Google Brain, and former chief scientist at Baidu. Today he spends most of his time as landing. His work has advanced the state of the art and machine learning research and he's taught millions of students artificial intelligence concepts online. We're going to dive into advice for businesses looking to leverage AI in their companies, how jobs will be affected by automation and the future of machine learning and AI. But first let's hear more about your career in research. Andrew, great to have you.

Thanks Sarah. Great to be here.

Walk us through your work and specifically the techniques in machine learning that you're researching. Really interested in applying today. I know you usually structured into supervised learning, unsupervised learning, reinforcement learning and transfer learning. So I think a lot of our listeners will have an understanding of machine learning in general. But can you talk a little bit about these modern techniques and how you're using them?

So AI today is creating tremendous economic value, but it's actually a large set of tools that we apply in different circumstances to different problems. So I think the vast majority of the economic value created by the modern rise of AI today is through one technique called supervised learning, which just means learning input to output mappings such as input an email and output is this spam or not? Or input an ad and output. Do you think the user will click on this or not? Or input a picture and tell me if your face is in this picture or input a picture, picture of what's in front of your car and tell me where are the positions of the other cars for a self driving car and so on and so on. And it turns out that when you can find the right business context to find these input to output mappings, this is called supervised learning. This turns out to be incredibly valuable. The other big categories of techniques that I think a lot of companies are finding very useful are transfer learning, unsupervised learning and reinforcement learning. So what is transfer learning? Let's say you want to read X ray images for a radiology problem, but you don't have that many X ray images, then it might be difficult to get a supervised learning algorithm to do well on it just because maybe you don't have enough data. Transfer learning is a technique that lets you learn from a huge data set of say, pictures of cats and people and pedestrians and cars, normal Terrestrial images, because you can get a lot of images of those and take the knowledge you learn from those millions of images and transfer it. So. So that when finally your AI system looks at a few X ray images, it can learn to diagnose, say, from X ray images, even from a much smaller data set. So this transfer learning, where you learn transfer from one task like recognizing cats and people and cows and dogs, to a different task like recognizing or diagnosing from X ray images, unsupervised learning is another technique that people are excited about. So that's having a computer look at a lot of data and just tell us, you know, hey, computer, look at this data and figure some stuff out. This is useful for a few applications. For, for example, today a lot of web search algorithms, or really algorithms that manipulate text, such as web search or chatbots, use one type of unsupervised learning algorithm that has read a ton of text on the Internet. And through just reading lots of text on the Internet without anyone kind of really telling you anything, has figured out, you know, a lot of things about the meaning of words. And through the understanding of meaning of words, it has learned off the Internet, it turns out that these AI algorithms then become much better at web search or chatbots and so on. And the final category, reinforcement learning, is a type of technique where the PR coverage, I think, is vastly greater than the actual instances or production products deployed using reinforcement learning. But reinforcement learning, I like to akin to how you would train a cat or how you train a dog, really, where you. You let the dog do whatever it does, and whenever it does something good, you go, oh, good dog. And whenever it does something bad, you go, oh, bad dog. And then over time, your dog figures out how to do more of the good things. So it turns out reinforcement learning is great for learning to play games like chess and go and checkers and so on, because whenever it wins a game, you go, oh, good computer. And when it loses, you go, bad computer. And then over time, it learns to play these games really well. So in terms of economic value created today, I would rank supervised learning, input output mappings as the first by far, followed by transfer learning, transferring from one task to another, followed by unsupervised learning. And then with reinforcement learning being really the distant loss in terms of economic value created today, having said that, the AI world is advancing rapidly, and so who knows, there might be breakthroughs, and this rank ordering might totally change the next two years.

So it's interesting, for a former academic, you rank these techniques by economic Value.

I think you're a vc, so I do that when talking to you. I do think about it in terms of the actual user value or just the impact on the world.

Right. So then in terms of impact on the world or effectiveness in terms of being able to solve new problems, are there other areas of machine learning research you think are really fertile right now?

I think there are a ton of exciting areas in machine learning. So just within deep learning, I think that recent advances in data synthesis through Gans generative adversarial networks, it's been amazing how accurately we can synthesize pictures of people in novel situations or synthesize audio clips of people saying stuff that they never actually said and, and I think certainly affect the entertainment industry. It might affect other industries as well, although I think that maybe remains to be seen. For many years there's been a lot of work on meta learning. There was actually very good work out of this at Harvard University several years back. And then my old team, the Google Brain team, is also doing a lot of work on this, but really using machine learning to speed up other machine learning algorithms. It's a cool idea, It's a very nice tactic. I don't know if it will follow fundamentally change everything, but it is a nice project to be working on. The AI community is so innovative and so creative. Many of these things you kind of see the seeds of it a few years earlier. The technical community is so global and we all talk to each other all around the world that the right ideas have the opportunity to catch on five very quickly and really go from zero to 100 in like a year. So this is an exciting time both to be doing basic research in AI and to be looking at all these great techniques and trying to figure out how we could use them to help people.

After Google, you led Chinese search giant Baidu's AI group of roughly 1300 people. You've talked about how you think AI is the new electricity. It can be used in as many use cases as you have the creativity to find. How do you decide what sort of problems to focus on at Baidu?

Yeah, AI is a new electricity. And like electricity 100 years ago, I think it would transform every industry. At Baidu, it's fortunate, fortunate to work with a fantastic team, including Robin Lee, the CEO, who's just a phenomenal executive, very smart about AI, but also many. The whole company up and down was just phenomenal in terms of developing technology and also figuring out new products. So the AI group that I led at Baidu worked both on supporting existing businesses. So we had many, many dozens of projects supporting existing businesses. Everything from web search to advertising to maps, to on and on and on to use AI to improve these different products. And then one thing that Baidu as a company I think was also very good at was to use AI to create brand new categories of products. So for example, when I was still there, Baidu launched the Duo s team, which provides a voice control platform for a lot of consumer electronic devices, for smart speakers and for other things. And it's actually in a very, in a strong position of leadership in China right now. So I guess the American version of that might be Amazon Alexa. Although one piece of history that's not widely known is that Baidu announced its smart speaker in public actually before Amazon announced the existence of Amazon Alexa. So it's very interesting how a lot of these ideas contemporaneously sometimes appear a bit earlier in the US sometimes appear a bit earlier in China. But I think at Baidu the AI group really two of the main activities were to support existing businesses and I think it helped many businesses as well as use the AI capabilities to start up new lines of businesses such as Duo S, the smart speaker voice control platform or the autonomous driving team from Baidu, which also came out of my team. When you look more broadly across society, across industries, I think that the opportunities for AI actually mirrors these two categories I just named at Baidu, which is that existing incumbents will have fantastic opportunities to transform using AI. That's actually what landing AI is focused on. But then there'll also be many new opportunities for brand new startups or maybe brand new industries to be created with this wave of new technology we now have access to.

Yeah, that's really exciting. And we should talk about landing AI and where you think, think some of those opportunities are that you're going after with the AI fund and other projects you're supporting. Just going back to what you were saying about Robin and Baidu, I think there are a lot of executives of tech and non tech companies in the United States that are very inspired by the transformation that Baidu went through to become a very sort of AI powered company as they describe it today. Do you think that the successes there were cultural? What do you think Robin and the Exec team team and obviously you and your leadership did to enable that change.

I think the Google Brain team really helped Google play the non trivial role in the transformation of Google into the great AI company a lot of people perceive it to be today. And I think the AI group at Baidu were responsible for the AI strategy and technology platform for Baidu, and that helped transform Baidu. But I think that for AI to reach its full potential, we need to transform not just great Internet companies, but really many companies in many different categories across many sectors of industry. I've been speaking with CEOs from multiple industries about recommendations I make to help really large incumbent companies transform for the age of AI. A couple of recommendations I tend to make. One is form an AI team that matrixes AI capabilities into multiple business units. If you're a CEO of a large enterprise and you have say five or ten business units reporting to you, it'll be very difficult to get all five or all 10 business unit leaders to build their own strong AI teams. So I think from a perspective of CEO, the odds of success will be higher if you can build an AI team that leverages both internal capabilities, build the things you need to build, but more importantly leverages external capabilities as as well. Well, don't buy the things that will become industry standard anyway or that an external party can develop much more efficiently and then matrices those capabilities into all of the different business units. So I think this is a playbook that will help many companies transform with AI. The tech world is evolving so fast. It's important that companies help your employees to keep learning because the tech will keep on shifting. Now, one other change over the last decade is that it's now possible to educate people much more conveniently and at lower cost than ever before. So 10 years ago, if you want to train people on AI, you know, well, what could you do, right? You could try to get me to show up and give a talk, or get you, Sarah, to show up and give a talk. But that isn't the most scalable way to help companies transform. But with the rise of online digital content, companies like Coursera, Coursera is seeing a ton of companies using this type of online digital content to very inexpensively train up huge numbers of employees on modern technologies.

That all makes sense in terms of strategy and channels to go bring that expertise in house and partner outside of it. Do you feel like there's a cultural difference in how Chinese and American executives see AI? I realize it's a leading question. I think there is, there is a.

Huge difference in culture both at the executive level as well as at all levels. The companies, I think between the US and China, a lot of companies in China are still younger, often founder led. And so on average, companies in China tend to make decisions very quickly, whereas some of the more mature, you know, Board controlled publicly traded companies. It's harder for say the CEO to very quickly change the organization to embrace a brand new technology direction. So that's one difference at the executive level. And, and really I think in China, so many people, you know, have seen fortunes made and seen fortunes lost all in the last 10 years that a lot of CEOs and chairpersons, they know that when the next technology wave comes, if you don't jump fast enough, it might be your turn to see your company go down in this wave. So, so I find willingness to adopt new technology very high in, in Asia. But having said that, I think the US has a stronger basic research and technology capability than China. So the US has at least today a significant advantage there. But I think continued investments seem like good investments really for many countries.

Let's talk about the AI Fund and some of the opportunities that you're working on now, like landing AI. We're really excited to be working on this with you, Andrew. Like why did you decide to start incubating companies and raise the fund?

With the rise of AI technologies, we can now make computers do things that no one could two or three years ago. In fact, I'm actually very inspired that today there are high school students that can download open source code and probably do stuff that the best research group in the world could not have done three years ago. What an exciting time to be alive now in terms of the opportunities this creates to help people, there are opportunities for incumbent companies to transform as well as to create brand new startups. So in fact, if you look at the last, maybe one of the last major ways of technological transformation, the rise of the Internet. The rise of the Internet allowed the creation of a few great startups like Google, Facebook, Baidu. Some of the incumbents also did a great job transforming. I think Apple and Microsoft did a great job transforming from a non Internet company to an Internet company. And so some of what I do actually tries to mirror that. Where landing AI is focused on helping incumbent companies transform for the age of AI, but also for AI to reach its full potential. There'll be many startups that I think need to be created to use these new abilities of machine learning and AI to do great new things. So the AI Fund is, is focus on that. The AI Fund's primary business activity is a bit different than most venture capitalists and our primary business activity is to initiate and to really build startups, AI powered startups from scratch. A lot of people have asked us if we're a startup incubator and we actually Say, no, we don't think of ourselves as an incubator. We think of ourselves as a startup initiator. But most commonly, we engage with entrepreneurs, AI engineers and various engineers and business people before their startup has been incorporated, before the team has joined, and sometimes even before they have a specific idea. And then we work with them to explore different ideas and then go from there to build the startups.

So one company that you've already initiated and you're actually serving as CEO of is Landing. What's the thesis of Landing and what kind of work are you doing now?

So landing AI is helping companies transform to adopt AI. And we decided to start in manufacturing. Before focusing on manufacturing, we did a systematic scan of multiple different industries, you know, including healthcare and education and a few others. And we thought that manufacturing is an industry where we can make inroads quite rapidly and help maybe as the first industry that will focus on transforming. Silicon Valley has done a great job transforming the digital universe. But really to your listeners, if your listeners look around, I think most of the pixels that hit their eye were actually created by the manufacturing industry. AI going to manufacturing. I'm excited about implementing a digital transformation of the physical universe around you. And I think maybe in some parts of the country, people have enough stuff, but if we could lower the cost of physical goods, what better way to give everyone in the world a 10% raise other than if we can make a lot of things 10% cheaper? So I'm excited about working with the manufacturing industry to improve access to everything from bicycles to drugs to washing machines. I spent a lot of time visiting factories, and I just find that exciting. But I think also we can help both the industry as well as the people that the manufacturing industry serves.

It's a massive vision. I feel like manufacturing has actually been a pretty underserved industry in terms of the penetration of AI or just modern digital technologies today. Many investors and entrepreneurs would say it's a segment of the economy that has not as strong a grasp of their own data as, for example, if you're an Internet business that understands your web analytics and your ad data and all of that, is that what you're seeing in, you know, in the market, talking to customers or like, are there interesting data sets to operate on here?

You know, so answering even more broadly than manufacturing, I think that I've actually yet to run across one industry vertical that is fully satisfied with its own data. So even look at the great AI companies, they do have better data infrastructure than many other companies, or maybe than all other companies. But I don't think even any of them are fully satisfied with the quality and the maturity or whatever of the data. One pattern I'm seeing across a lot of industries is that often first comes the IT revolution or the digitization revolution, and that creates data. And then a little bit behind the IT revolution comes the AI revolution, because AI can come and eat that data to create value. And over the last 20 years, so much of society has been going through the digitization revolution. For example, if you take an X ray, that X ray is now much more likely to be a digital image, like a computer picture, rather than a physical piece of X ray film. So that's data that AI can come and take to eat or take the supply chain, everything from X ray images to what you just ordered online. This is creating data in industry after industry is enabling AI to enter all of these industries.

And you're confident that that digitization step, that has to happen before AI really comes to eat the data? It's happened to a sufficient degree in manufacturing for you guys to go tackle the problem.

Yes, I'm spending a lot of time in manufacturing plans and no one's satisfied with quality data. Some of these plants, WI fi isn't up and running. Some of these machines aren't even hooked onto the Internet. And we're plugging USB drives into computers to carry data around. So it's just. So there's a huge variety and they're also highly automated factories where you walk to the plant and it's a giant plant and there are like 30 people in this huge building because it's all robots. So there's a huge range in terms of the degree to which different plants have been digitized. But one of the examples, exciting things, is also working with all this messy, dirty, incomplete data and sometimes even partnering with them to figure out how to make the data better. Some CEOs have come to me and said, hey, Andrew, this is what I'm going to do. Give me three years to build the perfect IT system, then I'll come back to you and I'll do AI then. And this is a terrible strategy, so just don't do that. Usually advice, if you've collected some data, it's probably worth getting AI team to take a look at it because. Because often the AI team can give feedback to your digitization team or your IT team on where to collect more data and maybe where to invest more and less.

That makes sense to figure out where you can get the value and then take a targeted approach to then collecting the data you do need. So before we move away from landing AI, give our listeners a little bit of color on one use case with whatever specifics you can offer value for manufacturing.

If you look at the great AI companies, you know, Google, Baidu, Facebook, Microsoft, Amazon and so on, AI isn't used for just one thing in these companies. It's not like it's online advertising and then it's done. It's actually used for dozens or maybe even more than dozens of businesses and the great AI companies. So it will be the same too in manufacturing companies. So some of the use cases in manufacturing include preventative maintenance and visual inspection. We can automate visual inspection and machine auto tuning to automatically optimize the yield or throughput of a single machine or of a line looking at this data that is not that easy to access otherwise, and then trying to develop products right around some of these needs to help our partners in manufacturing use their data better.

So in thinking about the broader impact of the adoption of some of these AI technologies, it's very inspiring that you and your team are focused on if we can make manufacturing more effective, more efficient, then obviously that should flow through to the consumers of goods. On the flip side, somebody in plants somewhere are doing visual inspection today, right? And I think it's pretty well accepted that application of AI technologies across a broad range of industries are going to lead to some jobs becoming less relevant than they used to be. How do you see, see the role of government or of companies in addressing the shift in jobs and skills that are relevant with the arrival of AI in all these industries, Governments have a.

Huge role to play with the rise of AI, both in addressing the job displacement and the workforce retraining. If we can be brutally honest, I think with the last few rounds of value creation, we did a great job creating a lot of value for parts of the country country, but really left a lot of people behind as well. And so if we're to create not just a wealthier society, but also a fairer one, then I think we have important work to do. I think retraining will be a large part of it. Today's economy is actually quite good at rewarding people with the skills needed to do important and valuable jobs. And in fact, it's not as if we're running out of jobs for people to do. We read these scary economics reports towards saying maybe in 20 years, 50% of jobs may be at cyclic risk of automation or something. The flip side of that is that maybe 50% of jobs are not at cyclical risk of automation. And in fact, here in the United States we can't seem to find enough teachers, we can't find enough healthcare workers, and we can't seem to find enough wind turbine technicians. So I think the problem is really one of re skilling or retraining workers to tackle the jobs that are very valuable in the future, in fact very valuable today. And people talk about computer programming a lot and clearly we don't have enough computer programmers. So I definitely encourage more people to enter that. But the solution is also not for everyone to become a computer programmer. But then there are these other jobs as well that are very much in demand. Silicon Valley has some excitement about unconditional basic income where you pay people to do whatever or do nothing. I like the intent behind that, but I don't think is actually the right solution. I think that is better. I actually support conditional basic income instead where we pay people to study because I think there's a lot to be said for the dignity of work. And rather than creating an engine that is more likely to trap people in low skill jobs with no career path out of like the gig economy, some of the gig economy jobs, I think it's actually better off if, if we use this type of conditional basic income to pay people to help them to study and thereby acquire the skills needed to reenter the workforce and contribute back to this tax base that could be paying for all of this. So there'll be innovative solutions like this that governments should develop even beyond our responsibility to citizens for making sure they have a shot at working and earning a great living for themselves and their families. I think governments worldwide are also waking up to varying degrees to the influence and the potential of AI Today. The US and China are ahead of probably all other countries in terms of AI technology and product and business product development. But we're still so early in the rise of AI that no one, not even the US and not even China, has a mature AI economy. So the pie of the economy and when talk about economic value or talk about value created or service decisions or whatever, but the value created by AI, that pie is small and rapidly growing. And I think any government that makes the right investments should aspire to help their citizens participate in this and maybe capture a meaningful slice of this. Still quite small, but I think rapidly growing pie. I'd be Quite sad if 20 years from now we look around the world and we say, yep, the US has done great and China has done great, but no other country managed to meaningfully develop its AI community or its AI economy. But the window of time to act is not unlimited. But I think for governments with basic services and engineering skill set, and some because AI is just so immature globally, I think there's an opportunity for all governments around the world to act now.

Spoken like a true former professor. The solution is to educate ourselves as employees or as governments on what the opportunity is ahead.

Yeah, but when you think about it, when we went from farming to manufacturing in the US economy and manufacturing to services, what did we do? We built the K12 educational system. We built a university educational system. And those educational systems, for all their flaws, for all their imperfections, changed generations of people's lives. And now we're seeing another technological wave. And I think the educational system needs to evolve as well.

It does require a lot of both courage and thoughtfulness on the part of our policymakers and people involved in making those decisions though. Because unlike us, where we in technology can sit and say like, how did that change our objective function? When you're experimenting with something like universal basic income, it's very hard to actually understand if the right incentives are in place or make decisions based on a 20 year cycle of like how does this actually impact people's place in society or work ethic or any of that. It's a complex problem to be thinking about.

Yeah. And one interesting cultural difference between the US and some of the countries is there are definitely countries, countries where government plays a much more active role than in the United States in really governing details of sometimes private citizens lives or development of the economy. So there are definitely countries where when a corporate leader has a problem, he or she picks up the phone and calls out the government and the government meets with them and tries to solve problems together. And on the flip side, when the government has a problem, they call up the CEOs of companies and then these CEOs actually help the government to solve whatever problem it has. So I've seen these public private partnerships sometimes go well, sometimes they waste resources. But this is actually one strength of some countries that is not part of the US right now. I don't think one model is necessarily better than the other. But to take the rise of AI as one example, I think countries where you can set up these public private partnerships, where government can work hand in hand with CEOs with new tech companies, they will be more efficient in adopting new technology like self driving cars. So there are some benefits to this as well.

Absolutely. I think a well known limiting factor for people developing self driving technologies in the United States today is their ability to test in different environments and get enough miles and enough data. And so we see a range of attitudes from local and state governments and a lot of people gravitating toward, for example, Arizona instead of California to do testing. And I think that may prove fruitful in terms of people building AI communities in places where we wouldn't have expected them.

Yeah, well, one interesting thing about the world today is that there are a lot of governments that are more thoughtful about how to partner with technology companies to adopt these technologies that will result in much better services being delivered to citizens. I think those systems of government will actually have an advantage in capturing a bigger piece for their citizens of this growing AI pie.

I think there's also a flip side to the cooperation that you just mentioned between really significant companies and governments in terms of in the US there's a lot of challenge put toward the AI enabled Internet companies about the consumer data they hold and the influence they have on broader society. And so I don't know that we're going to get to that topic in this podcast, but I think there is a question of as we collect more and more of this data, the US government, and supported by a lot of the US public has taken one tack toward ownership and usage of that data versus for example China.

And we have also the European privacy protection laws coming in place in May of 2018. I think that I have a point of view that might be unpopular in Silicon Valley.

Please share it.

Which is that it is actually a difficult question what rights a government should have to data about what is going on in its own country. So for example, what rights should the government have into getting data from ride sharing companies? It's actually a very difficult question because. Because arguably governments have a legitimate need to be able to see some data. Because if you don't even know what's going on in your own country, how can you govern effectively? Now the tricky part of that is we have to balance this against privacy of individuals. Most citizens in most countries are less worried about government surveillance than the US for some reason. So there is a large cultural difference between the US and other countries. And then the other thing that makes this debate tricky is that there are some governments that are quite oppressive to their own citizens and do not at all try to serve their own citizens. And then quite legitimately, individuals in tech companies might be reluctant to help non helpful, oppressive governments with providing even more data to them. But for governments that are well run, that have due process of law, and in the U.S. you know, we have a subpoena process, it actually works pretty well. Again, we like to try criticize it for its laws, but the fact that we criticize the subpoena process for its laws is actually a good sign. But I think it's actually a difficult question and not an easy one to answer. What are the rights? What are the legitimate rights that a government should have to understand what's going on in his own country?

We've already been talking about some suggestions you have for CEOs and other business executives about how to begin to to use AI in their companies and what sort of strategies they should put in place. And one of your suggestions was begin to build an AI team that can serve other business units and then think about how to transform the overall business and develop new products that are AI powered. One thing that we hear a lot from the companies executives in other industries is that the war for AI talent is feeding and all those people want to go work for you or for Google or for Facebook or for the next hot tech startup versus working in perhaps a traditional industry and helping to transform it. What would you say to them about the talent battle?

I think that we are fortunate that it is possible at lower cost and higher efficiency than ever before to train up your own native talent in AI. And then also, you know, there's this very poetic phrase in Chinese that would translate to English which is don't run in front of a train. And what that means is that if you are building an AI solution and your need is common in your industry, then there's actually a pretty decent chance that, you know, some neutral player will build an industry wide solution and can deploy much greater resources against that. Here's an example. Most companies should not try to build their own microchip. Some companies can try to do it, but most companies should just buy your silicon from Intel, Nvidia or Qualcomm or whoever. And that's because the industry wide solution, the chip makers have such greater economies of scale. So I think it's important for AI leaders, it's called the chief AI officer, the CIO or whatever, etc, to be very thoughtful about what to build and what to buy. And there will be many solutions in many industries where say a startup or a company can provide a industry wide solution much more efficiently than if every single company in that category tries to build its own solution. But of course if there is something that's completely unique to your business, then that is a good reason to build it. But having thoughtful chief AI officer or CIOs that can make that decision is actually really important.

Got it. So solutions which makes sense to me are use Coursera and then focus on areas of competitive advantage.

Use Coursera and whatever online content you could find, really, and find the right partners to bring in outside solutions when you have problems that are industry wide. And then either find the right partner or build internally only when you have of truly unique things. And maybe part of this is that most companies do not have the resources of Google or Baidu or Facebook or Amazon or Microsoft. The very large companies like that have large enough engineering workforces, they can build a lot of things by themselves. But we live in a world of limited resources. And so focusing those resources on what's really unique will be, I think, a key part of the AI strategy of companies.

So let's talk about AI strategy as it relates to me and you and some of our listeners that are entrepreneurs. When there are multiple talented teams focused on a particular industry or a problem set that requires AI, how do you think these companies will differentiate from each other? And like, we've talked a little bit specifically about how doing product management in an AI startup might be different than in a traditional quote unquote Internet company. But what do you think is going to differentiate those these companies? Is it the data? Is it the algorithms? Is it something else?

At least in this short window of time, AI technology is still really hard. And there's so many businesses where the accuracy of your AI technology makes a huge difference that I think a team with a strong AI technology capability already has a huge advantage. So one thing I'm seeing is there are actually a lot of entrepreneurs with lots of pretty decent ideas and hear about these business ideas. And I thought, yeah, that looks pretty good. But sometimes when you dig into the actual AI technology capability of their teams, you kind of feel like, boy, if only they had a stronger engineering team, they'd have a much better shot at this. So take an example for online advertising. Your ability to show an ad that someone is 5% chance of clicking on rather than 4% chance there's a direct impact on the bottom line. Or take speech recognition. The reason users are willing to use things like Amazon Echo, Amazon Alexa, or in China by Dideros is because speech recognition is actually very accurate. If Amazon Echo responded only half the time rather than 99% of the time or whatever it is, users won't accept it. So there are a lot of examples like this or the inspection work we do at landing the accuracy actually really matters. So I think today just having a strong AI team already gives a company a meaningful advantage.

People who can deliver sufficient accuracy for it to be useful and accepted.

Yeah. And accuracy is the one common way of measuring how good is your AI or how smart is your AI. And that is actually a business advantage today, when the AI talent is still so scarce and then longer term, I think that data is, is one defensible barrier to a business. There are multiple defensible barriers, but data is increasingly one more way of building a defensible business. And also strategy matters. These decisions we make on where to acquire data and which product to build and how to spot an opportunity to apply supervised learning, these decisions are still very complicated. On the DeepLearning AI specialization on Coursera, I actually do my honest best to teach people all these things and maybe despite my best efforts, it is still complicated.

Absolutely. So perhaps as a closing question, you're initiating a bunch of AI companies and yet still teaching deep learning AI and these online courses and making available the content required to become skilled in this area yourself. Why are you so interested in democratizing this?

I want to build an AI powered society. I think humanity is so much better off that we have electricity in most countries, not quite everywhere, but in a lot of places. And for AI to reach its full potential, I think we need millions of people able to use AI capabilities so that someday someone in a city or village may figure out a way to use AI to improve the water supply and improve the environment there. So there are so many applications of AI that no one company, not even the giant AI companies and not even Landing AI could possibly do all the work needed to create these things that are going to be really valuable for people. So with the DeepLearning AI set of courses on Coursera, I hope to give these tools to millions of people so they can then do this important work of transforming human society.

Wonderful. Well, it's been great to talk to you today, Andrew, and thanks for sharing your views with our listeners on the state of research in AI, some of the most interesting applications, what you're doing with the AI fund and Landing, and some broader implications of this on government and society. We wish you the best of luck with Landing. Talk to you soon.

Thanks, Sara.