Diálogos con Líderes Chinos del Futuro - Richard Tong. Inteligencia artificial y educación.

Yeah, so as you already talked about, AI is going to change every aspect of our life because of the power of automation and the power that a lot of the work that currently considered only human can do now can be in some way enhanced or in some cases even replaced by AI. And we see that AI is going to have similar impact in education, especially because traditionally, if you look at it, education is a very human-centric enterprise. Because you need a lot of interaction, also you need a lot of type of communication that is specific to human knowledge, you need to have really good understanding of the communication, emotions, and other aspects of the student in addition to the knowledge they're trying to acquire. However, there's one thing that the traditional education cannot do is to do what we call personalized education at scale. When we're looking at education, the natural way of education actually is one-on-one because when we are young, we get knowledge, we learn skills from our parents or the people who are around us. It's very interactive and also very personalized. You don't have three kids in the same room trying to learn from mommy or daddy at the same time, because each one of the persons here have different ways of communication, they have different emotional state, and they have different need at any time when they are learning things, especially life skills. But because in the traditional way, when the industrial revolution happened about two centuries ago, I think when started in Prussia, they needed some way to instill knowledge for industry, especially the skills like for military or industry as a worker or as a clerk. You need to have a lot of people who learn these things together and quickly, and that came with the more way of education which is like you have a classroom, a teacher sit in front or stand in front and teach everybody the same thing. But that type of education actually is very inefficient and also unnatural if you think about it. For every kid, even in a 30-student classroom, everybody has different pace, everyone has different aptitude and preferences, and also everyone has different challenges and struggles when they're learning stuff. But the problem with the factory-based model of today's classroom education is that the teacher is at the center of the education space, and the pace is determined by the average but not by individual need, and the interaction is minimal. It's mostly a broadcasting: the teacher is broadcasting knowledge to the student and trying to transfer to them, and the student is doing homework to get feedback later, and that feedback actually is not immediately available to the teacher to enable teaching. So the rich people or the privileged people typically hire tutors, right? That's why you see the aristocrats or the people who are really getting good education typically have a lot of people around them and give them very specific instruction and help. That is not doable with the traditional model in the new world to do personalized education because the cost is prohibitive. You need really good teachers and you have to dedicate their attention to each individual student, but that can be done by AI. So think about this: in the past, an Uber driver used to be that we have to take buses, right? So we go to places, we take bus, we take train because that is the economic way of doing transportation. But as the advance of GPS and also advances now with auto driving and so forth, we see a future where the cost of personalized transportation is getting cheaper and cheaper because so much more is done by automation and by scaling up the whole system. I serve you individually and help you to gain that efficiency through the automation part of it. So when we see in the future where the auto driving is available, we're going to see that cost is going to drop even further so that it's going to approach the cost of bus or trains which are mass transportation. And we see that happening in education as well. The more part of the education can be achieved through automation and through AI diagnosis, AI recommendation, AI feedback, the easier that we're going to achieve the personalized education at scale. So that is going to be the really important advances, and we see that as a paradigm shift in the education world, which is using AI to help each student to achieve their potential by giving them what is optimal at any time in real time, so that they will be like working with a personal tutor all the time. Even though they still need the teachers in the classroom to guide them and also have the emotional type of exchange and having disciplinary type of encouragement or motivational stuff, but I think more and more a lot of these teachers will be more like a head coach and the AI will be like a system coach, where the AI will be doing all the manual works, the tedious work, for example diagnose each question the student is answering and then give specific feedback on each of the student, assign assignment or task. And that will help the future where you can see the efficiency in education are exponentially improved, and then we can achieve a really high quality education at the cost, the similar marginal cost, and actually even cheaper than the traditional classroom teaching. Because once AI is getting the data and knowledge of the teaching and learning aspect, that cost is going to be the cost of electricity, and you can do 10,000 students at the same time, and you can do even more. If you look at like a Khan Academy or Coursera right now, they don't really have that much AI of customized learning, but just by making the content available at a large scale, that already reduces the cost of higher education, especially those educations like with community colleges or universities that doesn't have the top tier professors or instructors.

Absolutely. And this is actually I would say it should happen even before medicine, because the medicine and healthcare is probably more high stake than education, where you don't tolerate mistakes or you don't tolerate the way like for example when you do surgery or when you do even just prescribe a medicine, you actually need more control or more quality assurance. But in education, education is high stake but typically even you make certain mistakes like make a recommendation, it wastes a little bit of time but it's not life-threatening. So I think that is even easier for AI to be deployed. And another thing is about the data availability, because medicine is much more difficult to gain large quantity of data for each type of disease, for medical conditions. So for the AI model which requires a lot of data, a lot of iterations, it's actually more difficult for the medical world than in education. In education, we collect millions of students data and then billions of data points just in a single semester. So to us, it's a lot easier to train the model to give good AI advice. And then I think you see the real potential there. Because if you look at who need the personalized education and AI education the most, it's the people who are currently don't have even access to good teachers. Not only is it too costly for them, like in rural areas or in less developed economic conditions, they just don't have any teacher, not only a good teacher. A lot of them, if you look at the villages in China, some of the villages, the same teacher is teaching math, Chinese to multiple grades of kids at the same time. They're in the same classroom. For these kids, not only they're not different personally, but they are also not even the same age, right? So they need really personalized, differentiated education just because I need to learn like second grade math and the other kid in the same room need to learn fourth grade math, and you put this together and doing the same thing, it just doesn't make sense. But that's the reality. But with AI and 5G and also more ubiquitous mobile access to data, I think that is definitely not even a dream. We actually have deployed some of our solutions to the rural areas in China, where these are for charity purposes and also for experimental purposes. We have done that from our own charity effort, and also we have done that with Stanford University recently. Stanford actually ran a 10,000-student big experiment in Sichuan, and they have three groups: one group using AI, one group uses traditional learning technology like learning management system, and a third group using just traditional classroom. And we're still tabulating the data right now, but it's a resounding success from even day one that the kids have great results from even after a month. And this was again a six-month project, so we're still tabulating the data and running all the different type of analysis on that, but we see that it's going to happen. And then we see that as a real savior for the kids who are left behind and who are really struggling because of lack of human attention and quality teaching resources.

Yes, so we see that happening in any of the industrial revolutions, the four revolutions we had in the past, and that is already happening. As you can see, a lot of the insurance agents, a lot of the more manual working like even receptionist are losing their jobs because a lot of these things are automated for them and the AI is taking over. For example, just receiving phone calls. When you call United Airlines or anyone, the first you receive is a robotic voice and they're going to tell you what you want and route you if necessary to a human recipient of the phone call. And that is going to happen no matter what because that's the way technology advances. In the past, this is similar to when the horsemen or the people who were the cart coaches, the person who drives the dragons, and they take care of the horses, they basically take care of the transportation need using traditional way, and they were replaced when people get personal cars. These people lose their jobs, and that's what happened because now everybody once they can drive themselves, they don't typically need somebody who can take a cart or a wagon. So what I'm seeing though is that as things disappearing, the need also changes and created a lot more opportunities. Because if you think about it, even the AI can take care of 95% of the education need for the kids, there are still very important human touch that is needed. Like we said, we need head coaches. In the past, it's not because it's not needed, it's because the teacher who didn't have time to take care of this individual need and having the attention for each and every kid. They typically have maybe a little bit of time for the really good kid or maybe one or two of the really slow kid, and that's pretty much it from their time perspective. But now if AI takes over the manual part of it, I see the teacher actually are more needed than before. And also they need new skills, they need to be able to work with AI in a very productive way, they need to be able to understand what the statistics, what are the AI recommendations means, and also they can help to explain and enhance AI in corner situations that are not mainstream. Because as AI is learning, you see that immediately they learn the basic stuff, the thing that can be repeated again and again. But for each case, there are few corner things that AI just miss or they need further human instruction or supervision. We have this paradigm called human in the loop. It's actually very important that we keep human in the loop in the AI revolution. It's not that we're replacing the teacher, it's that we're reducing the unnecessary tedious and boring workload that they need to deal with today, and giving them more innovation space, giving them more time to connect emotionally and also even morally with the student, so that they can be the driving force behind a lot of the AIs. We call that continuous learning and continuous improvement for both AI and the human. And you can see that for doctors, I don't know whether we heard the machine called Da Vinci, which is a surgery robot. With the advance of that, the robot can do a lot of the high precision surgeries with a human somehow but also with a lot of automation, a lot of monitoring and so on. You see that because of the capabilities available now, there is a higher demand for surgeons, not a reduced amount of surgeon, because now a lot of things that were not possible before become possible. And then you actually need more doctors, and the doctors can take shorter time to become really good at delivering high quality medical services. And I see the same thing happening: you will see these super teachers that are really enhanced by the AI and doing things that were not possible, like individualized diagnosis, individualized attention, with a lot of the AI automation, and they can be so much more powerful and be able to delivering things at the cost that is still reasonable or in line with the current education cost.

Thank you very much, really appreciate it. Bye.