Vinod Baktavachalam, Coursera - on online education in Russia, student motivation with the help of machine learning and gender stereotypes

Courserais an international online education platform with more than 37 million registered users.

It offers about 2,900 courses, 300specializations, there are separate courses and specializations, the highest level is degrees (higher education according to European standards). Educational content for courses is created by 160 universities and other educational institutions, as well as 30 industrial partners.

“Russians are more focused than others on the application of what they studied in their work”

- You recently conducted a study on the skills of Russian students for the first time. What is the specificity of Russian online students?

- We have more than 940 thousand. registered users from Russia, this is the second largest market in Europe after the UK and the largest non-English-speaking market. It grows very fast - 40% each year. Involvement is also very high - users from Russia have subscribed to 3.5 million courses, that is, about four courses per person, as many as in the United States or in the United Kingdom.

- What about gender and age?

- The average age is about the same as onglobal level. But Russian users are more balanced in gender terms - 47% of women, 53% of men, in the world 42% against 58%. This is generally consistent with what we see in the United States and England, and suggests that in Russia everyone is interested in high-quality education and is very motivated to study. Most Russians are interested in data analysis, business and computer science. They want to build a career - about 67% voiced such a goal. They seek to develop in the current position or get promoted, switch to something new and acquire new skills. This is above the general level of 60%, that is, Russians are more focused than others on the application of what they studied in their work.

Photo: Anton Karliner / Haytek

- And they achieve these goals?

— Yes, we track long-term resultstraining all users on our platform within six months after the person completed the course. We ask if the goal has been achieved. In 85% of cases, users from Russia who indicated a career report achieving this goal, 35% received a promotion, found a new job or opened their own business. This is a very strong correlation between the goal set and the goal achieved.

Sberbank and Yandex are looking for employees using content

- It seems to me that Russian companies do not take online certificates seriously.

- Depends on the company.Yes, we see this in the USA too: there is also a big gap between progressive companies and everyone else. In Russia, leading technology companies - Sberbank (Corporate University) and Yandex - create content on Coursera, invest in online training for their employees, and are looking for new employees among those who have taken courses and received certificates. Such innovative companies do not see the difference between online and offline education, they simply need people with the required skills, with high motivation to learn new things, since technologies are constantly changing.

Only large companies can afford thisbecause they have the resources, they can devote more time to the employee's education. Small companies still consider online training risky; it is much easier for them to hire people with a university degree, because it is clearer how to select high-level candidates. But more and more companies are realizing that hiring people with specific skills is not easy, and online education is a cheap way to retrain existing employees.

- And which companies use this opportunity?

— Google has a basic level certificate forall IT positions - computer technicians, network specialists and others. Not only have they created content, but they are also building hiring partnerships so that people with this certification can apply for jobs at Google, Walmart, and other companies.

- Do you do some customization of user experience depending on the country?

— The difference between countries is small, muchUsers differ more strongly at the individual level—they have different skill sets, different life experiences, and different goals. And we are in the process of personalizing the user experience on an individual level, using machine learning. But in terms of building partnerships, the Russian market is very different from the American market.

- And what is the difference between your business partners in Russia and in the USA?

- From my point of view, Russian universities andindustry partners are much more willing to work together, they are more innovative, they want to try new things more. We cooperate with HSE, MIPT, Yandex, Sberbank. Together they create both academic content and practical industry courses. HSE and Yandex, for example, made a theoretical educational program of high level in machine learning, and within the program a course on how to win a machine learning competition: tricks and tricks from the winners of such competitions, that is, direct experience in the processing of theoretical knowledge specialists . This is a unique situation, peculiar only to Russia, in other countries we have no examples of such partnership. Russian companies are expanding their horizons in terms of hiring for key skills - the same Sberbank and Yandex are looking for people using content.

Online training will come with fellow students

- And inside the course, how do you use machine learning to keep students motivated?

— There are two situations where we usepersonalization. The first is to search for new courses; they should be as relevant to the user as possible. This is where big data comes to the rescue. We're building a series of algorithms to connect skills with content that helps them acquire them, with career opportunities that require those skills, and with users with those skills. Thanks to analysis, a person finds the courses that best suit him.

Photo: Anton Karliner / Haytek

The second point is support inside the course whenUser has encountered difficult stuff. This is where the “smart tutor” turns on, which sends customized messages. For example, when you failed the test and lost motivation, we write: “Do not worry, 70% of people did not pass this task on the first attempt, here are links to materials that helped other users in this situation.” At the beginning of the course, we can write: “68% of those who completed this course received a promotion or a new job.” Every week we slightly push the user, and thanks to this he ends the course. Gradually, we can figure out which types of messages work well for you, and use them further.

- For some reason, I was offered courses on machine learning, although I am interested in psychology, and I had already signed up for a course in psychology several years ago.

- Yes, this is a problem. We have about 3 thousand.courses on a variety of topics. Tens of millions of users. If you multiply all this together, you get billions of recommendations. Personalization is very difficult. We started by recommending the selections of other people who had joined the same course. It worked successfully, but not with new users, and any person who has not done anything on the platform for a long time is new to the system. We use data from the last six months because people's preferences change greatly over time. So we decided to just ask people directly: what are your goals? Then we learn about your experience in this industry, and finally we make a recommendation.

Best Coursera on Big Data Analysis by Vinod Baktavachalam

Building a business strategy in a company based on big data analysis:

  • Econometrics.
  • A Crash Course in Causality: Inferring Causal Effects from Observational Data
  • Econometrics
  • Causal Inference

Learning machine learning theory to create products based on it:

  • Machine learning
  • Advanced Machine Learning Specialization
  • Deep learning specialization
  • Machine learning specialization

— What main trends in online education do you see?

- First - the level of personalization will grow. We have collected a lot of data about users: what they learn and how, at what time of the day, whether they listen for ten hours in a row or 15 minutes every day. Online experience is very similar to classroom training. This is just a set of videos: you sit and watch them, one after another. In the future, navigation through the course will be more free: first, you will pass the test, so that we understand which part of the course you are sent to, then if something is too difficult, you will be taken a step back, then will be offered a test of knowledge of the material. If this is too easy, you can skip part of the educational program.

The second trend is cheap technology scaling. The advantage of learning in the classroom in collaboration with fellow students. If we transfer this to online training, everything will change a lot. We still do not know how to synchronously connect millions of people. So far, it works for a very small number of people — through forums, web conferences integrated into courses.

- What about augmented and virtual reality?

- These technologies are very relevant forindustrial training. But I can imagine how we will virtually interact with colleagues on solving business problems. If we succeed in reproducing the working environment very precisely for practical skills training, we will be able to compete with universities. This is a question for the distant future. We often discuss at conferences that online learning cannot provide communication skills: how to work in a team or organize working time. That's just the right way for such skills VR - for example, to simulate the experience of interaction with consumers.

First year of college spend online

- What will go online soon?

- Technical directions - data analysis, infirst of all. He has so many requests that universities simply cannot satisfy them. Online you can get feedback, study cases and many other advantages.

- What about more traditional specialties?

— There is potential in other areas too.Perhaps not in the near future, because universities are very interested in preserving offline education. But you can safely transfer the first year of study at any college online - all these introductions to biology and chemistry. Some institutions are already doing this. Thus education will be broken into parts. After studying the theory online, students in classrooms will be immersed in the specialty. In addition, this is a way to reduce the cost of education, which is constantly rising.

- What are the other disadvantages of online education?

- Many say that the university is morelearned from each other, working in groups than in lectures. If we can move it online, it’s a powerful tool. The second disadvantage is the lack of networking. At university, you communicate with people with similar views, they, in turn, know someone who will help you get a job. Online, of course, you can find a company that wants to hire you for your skills. But most companies want people with recommendations.

Adult story

- What are you working on now? What major changes will Coursera expect in the near future?

— We are working on a comparative analysis of skills.Using this option, you can very effectively assess the level of training, build a personal user profile, and offer candidates to companies that are already on the platform. For example, some company needs a digital marketer. We send her a profile of the ideal digital marketer. And for those who want to become a digital marketer, we offer a skill diagnostic, compare their profile with the ideal option, note strengths and weaknesses, and suggest suitable courses and the order in which they should be completed. Once a person reaches the required level, we send a message: You are now qualified enough to work for such and such companies, and you are even better than some of the employees at these companies who also use Coursera.

Photo: Anton Karliner / Haytek

— Are you planning to create courses for children and teenagers?

— Our specialty is adults.With 50% of the world's jobs at risk of automation, there is a great need to retrain mid- and late-career workers. We have already built an ecosystem, and this is our expertise.

- You have an interesting study on the effect of population aging on economic growth. Share it with the results?

- Yes, in the economic literature there are twohypotheses. At first it seems: due to aging, growth slows down, because people of working age become smaller. But there are other studies that say: if there is less labor, then it is more difficult for companies to find workers, and they have to adapt, replace people with technologies, thanks to which the economy does not slow down. I tried to analyze these two hypotheses and came to the conclusion that companies are not effectively studying demographic trends, do not invest in the distant, in their opinion, future. Some adapt, but how? Are they replacing labor or just using technology to make employees more efficient? There is both. Computers automate the assembly line, collectors lose their jobs, but there is a need for technicians who will create and maintain robots. Just need a new skill set.

- And what are the strangest courses on Coursera?

- We have all. For example, the fake news course: why there are so many of them, how to tell lies from good news.

- And the most popular?

- Everything related to business, computer science anddata analysis. Now the most popular courses are about machine learning, artificial intelligence and especially about deep learning. Specific data analysis tools are in demand: cloud technologies, platforms offer courses, how to use their infrastructure. Companies often train their employees through such courses. Also popular are courses on open source technologies, for example, TensorFlow from Google. Popular courses on bitcoin.

- Still?

- Yes, funny, there is a direct correlation betweenbitcoin course and interest in bitcoin courses. In general, the blockchain is still a very high trend. Many companies want to hire blockchain-developers, but universities do not teach it yet. We have industry partners who have created very successful training products.

- How do you build ethical questions?

- We want to build a platform on whichuniversities studying ethics, together with companies that make money with technology, will work on these issues - what to do with artificial intelligence and so on. In general, we constantly discuss issues of ethics in the company. For example, initially we had a separation by gender in the recommendations, and yes (what a surprise!) - women choose engineering specialties less often. But we do not want to perpetuate all the same stereotypes as offline education, so we turned off this option. The moral compass is very important.