Andrew Un's next trick: to train a million experts on AI

Lead AI researcher and teacher claims that millions of people need to understand deep learning

Andrew Eun, one of the world's most famous experts in the field of AI, is launching an online project that should help create a million new experts on AI in various fields of production. Eun, one of the pioneers of online learning, hopes that his new course in depth education on the Coursera will teach people to use the most powerful idea from the field of AI in the last few years.



Experts in AI have become the most sought-after and well-paid employees in the modern technology industry. In-depth training is a machine learning to perform complex tasks using large amounts of data and a large neural network simulation. For the qualitative development and application of this technology usually requires deep technical knowledge and experience.



In March of this year, Mr. Yun resigned from his post as chief science specialist from the Chinese Internet company Baidu, in connection with which various gossip about his future plans began. He says that the in-depth training course is the first of three projects that he plans to launch in his new startup, Deeplearning.ai.



Before joining Baidu, Eun was a professor at Stanford, and then the founder of Coursera, and created a machine learning course that brought together more than two million students over the past few years. A few years ago, UN became the director and founder of the Google Brain project, aimed at introducing in-depth training throughout the company.



He spoke with the editor of MIT Technology Review about his mission to change the world through in-depth training.

Why did you choose to launch this course as your new big project?

What interests me most of all today is the construction of a new society working with the help of AI. Although the main hype associated with AI, refers to large technology companies - and it is obvious that such companies create tremendous value with the help of AI, improving search in the web, online advertising, cards, payment systems, etc. - If you look at the economy as a whole, then any Fortune 500 company can also create tremendous value with AI.



I don’t think that any one company will be able to do all the work, so I believe that the only way to achieve results is to teach millions of people to use these AI tools so that they themselves can invent such things that no large company or I built company can not do.



I think the world will be better if the AI ​​helps us. He will reduce the cost of goods, give us a good education, change the system of work of our hospitals and health care - yes, a lot.

Do we really need so many experts?

I do not think that every inhabitant of the planet should study in-depth training. But if AI is a new electricity, then look at how many engineers and electricians there are. In order for society to understand how to create the many wonderful things that are available to us, it needs to create a huge work force. So today labor in AI is much less than required.



Although the system of university education is gradually shifting towards AI, we cannot rely on universities alone. When I learned how to program 25 years ago, I learned BASIC, then the world changed, I learned C, then C ++, then Java, then Python. And I think that today the world is changing even faster. I think that the change under the influence of in-depth training is even more significant, therefore the creation of values ​​and the need for new skills is very obvious.



I talked to the directors of very large non-technology companies, and we came up with ideas for new AI products suitable for use in their industry, but they don’t have people with GO skills who could create these products. If you control tens of thousands of people and you don’t have an AI group, what can you do? I know that some of the directors are going to send their employees for training on the GO specialization.

How do your attempts characterize existing academic organizations?

I think you need to learn all your life. In the technological world, we are used to it. Something new appears every few weeks. We are used to the constant movement of sand beneath our feet, and I like it. It seems to me that this makes the work in the technology industry more interesting. But the revolution of AI and GO is uniquely fast. And although universities are increasing their learning opportunities, too many people who have already left universities should study these new systems.



I want to create a society that lives thanks to AI, because I see so many ways that AI can improve human life. We can make decisions more systematically, or automate repetitive tasks and save human time. It seems to me that there are so many good opportunities, and so few people with the skills necessary for their implementation.

Can GO help automate the learning process itself?

I hope that GO will play a role in changing the education system. Even think it is obvious. Education is one example of an industry with great potential for AI. And Coursera is already doing something in this direction.



Coursera holds tons of educational data. We know which videos students watch, in which issues they are mistaken. Coursera has the opportunity to study this data, and we are already working in this direction.



In many industries there are two waves of transformation. First it’s an IT transformation, and then an AI transformation comes. I think that education has a lot to strive for in terms of IT, but there are many opportunities for AI in this area. I recently attended an educational conference and was surprised at how many reports were related to AI.

Can we call this part of an attempt to compensate for the possible negative consequences of introducing AI?

The strongest negative consequence of the introduction of AI will be the disappearance of jobs, and this can not be avoided. But, creating a problem, we must work on its solution, and this solution will be the improvement of education. The society will not finish the work; just for new jobs, today's skills do not fit. I think this is an education problem. And we will need something more than just a Coursera - we need a public rethinking of the education system.



"Elitist" from the West and East coasts made the West and East coasts richer. But we need to create not only a richer, but also a more just society, much needs to be done in terms of changing the social contract. In this puzzle, more pieces. One of them is just to bring many more people to the GO region.

What do you think about China’s recently announced plans to dominate AI?

In various areas of the Chinese industry there is a huge interest in AI, and I’m very worried about reducing funding for science and technology in the US. About 10 years ago, my work at Stanford was financed by the National Science Association and DARPA, and without this I could not have done all that I could. It also allowed me to propose the Brain project to Google.



The question is which of the young professors will not receive funding, and will not be able to do anything in five years because the NNA did not have enough money for their proposals.



It scares me. Conversely, other countries, for example, China, are actively investing in the study of AI, and their environment makes it easier to do this work. With them, I see a great desire to apply new technologies.