"We are living through an era where AI research output is being released at an unprecedented rate and volume"

Gereltuya Bayanmunkh is a 27 years old student of “Artificial Intelligence and Data Science” programme at St Petersburg University from Ulaanbaatar, the capital city of Mongolia. Before coming to St Petersburg she studied at the National University of Mongolia for her undergraduate degree. She shared with us her experience of living in St Petersburg, her learning path and view on the future of specialists in AI.

What can you tell us about your experience of life in St Petersburg?

I was born in Ulaanbaatar, the capital city of Mongolia, but growing up, I lived in different provinces of Mongolia, including Uvs, Khovd, and Uvurkhangai.Compared to living in Ulaanbaatar, my home town, living in Saint Petersburg differs greatly on few points. Firstly, Saint Petersburg has 4 times more people living in it than my city of birth. The big city energy here is exciting. Secondly, the climate is much better and milder in winter here, which is a pleasant surprise. Thirdly, I really love the fact that the art scene in Saint Petersburg is very active in both traditional and modern sense. But compared to Kazan, in which I lived previously, my dormitory here has fewer foreign students, so I would like to invite more and more foreign students to study here in this beautiful city.

How long did you study Russian? Where did you start learning Russian?

I have been learning Russian for the last 2.5 years now. Firstly I learned the very basics of the language, up until the level A2, at the Russian Center of Science and Culture in Ulaanbaatar (now with the name The Russian House in Ulaanbaatar) in Mongolia in the autumn of 2020. Then I continued my language education, up until the level B1, at the preparatory faculty of the Kazan Federal University for the academic year of 2021-2022.

What major did you choose while earning a bachelor degree and why?

I studied Mathematics for my bachelor degree at the National University of Mongolia. Ever since I was a little girl, I always loved Mathematics, and spent most of my free time involved in its studies to an extent. Although the subject has become bigger and more complex as I grew older, I found Mathematics to be a beautiful, orderly, and confident science. After finishing high school and studying Information Systems for a couple of years, I wanted to learn more about how Mathematics handles more abstract concepts, how it is applied to other sciences, and find out which subfields of Mathematics I would be more drawn to.

Why did you choose SPbU to continue your education?

When I was initially planning my graduate studies in 2020 and researching Russian universities, I found out that St Petersburg University has a great Mathematics and Mechanics faculty. Then I learned that students can study Computer Science-related subjects along with Applied Mathematics ones. Since I was looking for AI programmes with a strong theoretical foundation, St Petersburg University felt like a great choice for me.

What inspired you to start studying AI?

Still being a student at my previous university, I attended a summer school in deep learning in my country 5 years ago. Then, I was working as a web developer (since I have a bit of software engineering experience) and I was looking for a way to study more about AI. The abovementioned deep learning summer school was the first of its kind in Mongolia, and it was an eye-opening experience that solidified my career decision to study AI more in depth and pursue a career in the field. The first structure I used seriously was RNN (Recurrent neural network), and I used it to train a deep learning model that is fed by the full texts of Mongolian laws in force then to generate text resembling a written law in 2018. The training took around 26 hours on a normal machine with CPU and the first parts it learned to generate were law structures with enumeration, section titles, and signature on the bottom. This is a simple result by any means, but it blew my mind at the time, and left me wondering what future AI can do!

What is the most difficult in studying ‘Artificial Intelligence and Big Data Science’ programme in Russian?

As I am still learning the Russian language at levels of B1-B2, the language is definitely the hardest part of my studies for now. But since our professors are supportive, and most of them are good English speakers, I sometimes turn to English communication as a last resort, and I am very grateful for this possibility.

When did you first come to know of Artificial Intelligence? Has your opinion changed since then or not?

The first time I came to know of AI was around 2016-2017. I did not know specifically about the field or the scientific aspects of it, but I believed it to be an Ikigai spot (or a career sweet spot) in my career. Now, as I learn more and more about AI as a scientific discipline and a field that brews great tools and techniques for real-life applications, my belief has strengthened, fortunately.

In which area do you want to apply your knowledge and skills in the future?

In the future, I would like to apply my knowledge in the field of AI and Big data science for the social good in an ethical way. Specifically, I would like to learn more about how AI techniques and tools can be used to fight climate change and help us adapt to the impacts of climate change more effectively.

What academic achievements do you hope to accomplish while studying?

During my master’s studies, I would like to write and publish an academic paper in the emerging interdisciplinary field of Climate change AI. I hope to participate in a scientific conference of the same field too.

What subject of your major do you find the most interesting?

Out of the courses we took during the first term this academic year, I found the Fundamentals of Machine Learning and the Theory of Bayesian Networks to be the most interesting. Also, I took a course named Intelligent Computer Vision Technologies mistakenly, due to a schedule misunderstanding on my part, and found the course content to be very interesting.

If you were asked to write a short article about Artificial Intelligence, what would you write?

This question is very relatable, because just yesterday, I published a short article on my personal blog about how the descriptions of the term Artificial Intelligence changed over the years, since the term was first coined in 1956. Now I am planning to write about the timeline of innovations in the AI subfield Natural language processing, and how those innovations influenced the creation, and the current publicity of the Large language models and Generative AI models.

What skills and knowledge do you consider the most important in your area of study?

Right now, we are living through an era where AI research output is being released at an unprecedented rate and volume, and AI innovation is being deployed in different industries every single day. To study and practice in the field of Artificial Intelligence in this rich environment, one has to learn how to sort through and process large amounts of information effectively, and stay focused on the given task at hand. Training large models at scale with so much resources is cool, and is important to the progress of the scientific field and its industry innovations; but from the start, we have to be conscious of the ethical implications of the resulting AI models and use cases, and of the environmental impact of training them and maintaining the use cases.

What kinds of projects using Artificial intelligence technology are interesting to you?

I am fascinated by the innovations made in the Natural language processing space, especially since 2017, so projects that process and understand text data using AI are very interesting to me. For example, I am now using a Contextualized topic modeling tool for my practice project with the help of my scientific advisor Dr. Dmitry Alekseevich Grigoriev, and the techniques used to identify topics within large text data are very promising. Recently I read about how an AI model is being used to process and generate music based on text prompts, and I found this project type to be also very interesting.

What difficulties did you have to face while studying?

To process scientific materials, including journal articles and book chapters, very quickly and gain enough insight to refer to it later on with confidence and clarity is quite difficult for me. I also find it hard to stay focused when researching about a certain topic and tend to fall into a rabbit hole of related topics. As a master’s student, planning my own studies effectively and allocating time accurately for certain research or practice tasks are also some points I am willing to improve upon during the length of my remaining studies.

How do you think in which areas will Artificial Intelligence enable more growth in the future?

I’ve read that multimodal neural networks are being actively developed at leading research institutions around the globe. This means a single AI model or, at least, a single front of multiple AI models will be able to process different types of data all at once and output appropriate results. This, along with powerful reinforcement learning models and structures that learn from the environment without much human supervision, is something I look forward to in the near future of AI. As for the implementation, I think that today’s generative AI models will take its shape in a more profound way and get deployed in technologies that help humans live more creatively and purposefully.

In your opinion, how long does it take to become a great specialist in this field?

Today’s leading researchers and practitioners have dedicated their entire careers to studying, and moving the Artificial Intelligence field forward. Also, the AI field is developing at a very fast pace, so even if one is a specialist in the current techniques, it does not guarantee that he or she would stay as a specialist in the future. But I believe that if one dedicates at least 5 years for formal or self-education of AI, and real-world practice, then he or she can keep pace with the AI world and become a great specialist in the direction of one’s choice further on.

How do you see your future profession? What are you going to do after graduation?

As for my profession, firstly I would like to stay up to date with the current and future developments in the Climate change AI subfield. Secondly, I would like to contribute to the global goals of reaching net-zero greenhouse gas emissions by 2050 with whatever I can do. This means that after graduation, I will probably be working at an organization that helps my country, Mongolia, to adapt to the impacts of climate change with the help of AI technology in some ways.