.png)
The new wave in the development of artificial intelligence and machine learning has made them much more accessible. How can they help business operations and in academia, how are they changing our education system and the labor market, how will they affect the data analysis industry, and why is the demand for intermediaries growing? These questions were discussed at the NES Popular Science Days by Ivan Stelmakh, NES Professor, and the School’s graduates, Daria Levina, Strategy and Operations Manager at DataGPT (Canada), and Stanislav Mazurenko, Team Lead, AI in Protein Engineering at Masaryk University (Czech Republic). GURU summarizes the main arguments of the discussion.
Watch the video (in Russian)
What Has Artificial Intelligence Changed
Ivan Stelmakh: Two important events took place in Russia in 2023. Firstly, Yandex has released its own version of the [generative AI model] GPT – YandexGPT. It is already being implemented in certain products. This fact proves that Russia is still competitive, it is not far behind the global market and has the technology that can change business operations and life. Secondly, Tinkoff Bank and its partners began to build their own university. Large companies with businesses largely based on AI technologies feel the need for faster intellectual turnover and training of people who would develop AI in Russia.
Daria Levina: I was struck by the unprecedented speed with which AI penetrated into various processes over the last year. It is actively used in literally all areas of online business. My personal benefit from using AI technologies is the increase in my productivity as a data analyst, including, among other things, translation of data into text format. The amount of information that can be processed using AI has increased 3-5 times.
Stanislav Mazurenko: In biology, AI has been used for a long time in small models that were trained on a limited amount of data. In the last decade, interest in this topic has increased multiple times. Experimental technologies have emerged that allow us to collect thousands of observations in a couple of weeks, whereas it took years before. In protein engineering, a breakthrough occurred about five years ago. It was associated with the prediction of protein structure, a field in which scientists have competed since the 1990s. In 2018 Google DeepMind, an AI research laboratory, released the first, and two years later, a more accurate second version of the AlphaFold algorithm for predicting protein structure. This invention came as a shock to the entire structural biology industry.
AI opens great opportunities in fundamental science. All biological systems are very complex and interconnected, so their interaction is difficult to analyze using classical methods. AI allows us to automate processes. Biology as a science often uses tools from related disciplines: for example, it draws inspiration from computer vision technologies and large language models. My personal benefit from the advent of AI technology is that my work has become possible. More and more people are looking at AI and wondering: can we apply it in our field? Before, being able to use it required specific knowledge, now this tool has become more accessible.
Ivan: What has changed from a practical point of view? On the one hand, everyone has now become an AI expert, researcher and startup founder. Everyone has access to technology, and it is very simple: people who have never coded before, now participate in hackathons where you need to create something based on AI. To do this, you just need to come up with the right ChatGPT query and nicely arrange the result into a presentation. We see more creativity around this field, which it lacked before.
On the other hand, a logical question arises: how can we use machine learning algorithms to solve social or economic problems? In September 2023, a paper was published. Its authors studied whether an algorithm predicting which of the students would not be able to complete their studies was useful for schools. It turned out that the model made very accurate predictions, but its use actually did not lower the share of expulsion. We see that machine learning tools are developing, but it does not always help people solve practical problems. Therefore, the key question is: how can we properly set up the transfer of technologies described in academic research or experiments to solve real problems of firms, economy or society?
How the Data Analytics Industry is Developing
Daria: The main task now is to increase the availability of data for users. Before, a data analyst needed to know programming languages – SQL and Python. Now there are many business intelligence tools, including Power BI, Tableau, etc., that allow people to enter a query into the program's dialog box, and it will visualize data, provide information or an explanation of the changes. These tools assume that you know or understand the question you are asking. Let's say you have a hypothesis: sales fell because of a certain category. Then you can ask: I want to see sales dynamics by categories for a certain period.
The product that we are currently developing at DataGPT is the next stage in the development of technology, when a person will no longer even need to know what he or she is looking for. It consists of two parts. One part analyzes in full all changes in the data and looks at what factors led to revenue growth, traffic drop, etc. A person does not even need to formulate a hypothesis – the data has already been analyzed, so you can simply ask a question in text format and get an answer in the same format. For example, you can ask: "Why did revenue fall?" Anyone who has nothing to do with data analytics can use such an instrument.
Our technology can analyze the data and say that revenue, for example, has changed because certain parameters have changed. But only a person who understands the company's business can understand why they have changed. And only this person will be able to take the research to the next, more advanced level. For example, run complex tests and see how the metric changes when one parameter is altered. Before, people simply did not have enough time for this, and now they will thanks to handing over their routine tasks to AI.
My experience of working with firms shows that the introduction of AI tools and just real-time data monitoring always have a positive impact on revenue. First of all, this is true for online businesses: they have a lot of data on user behavior. New features allow, for example, to quickly identify and eliminate specific inefficiencies. Technology increases firms’ reaction speed: what took three days to figure out the causes, now takes minutes.
Firms invest in the development of data analytics: they hire specialists, purchase software solutions, but at the same time the share of organizations that can call themselves data-oriented does not grow and remains at around 25%. That is, despite investments in this area, insights that are derived from data cannot always be integrated into real life.
What Prevents the Transition to New Technologies
Stanislav: There are universal and industry-specific barriers. The former include the lack of professionals with a background in computer science, computational methods, and data management. Not only do you need to know how to train a model, but you also need to be able to determine where the data comes from and what information to extract from it. Science operates with complex data, so people need to be trained to use it meaningfully.
Anyway, I think that it is easier to organize a technological transition in academia than in business operations. Firstly, knowledge and models are freely available. Thus, researchers do not need to develop solutions from scratch, they can just adapt existing ones. Secondly, scholars are curious people, they don't need to convince employees to try new technologies. A specific barrier in my field is the validation of new models, which takes time.
Daria: I think that AI will sooner or later be used in all business tasks. However, changes are often a source of psychological discomfort for people, especially if they are used to the existing technology, in which they have invested a lot of effort and which they adjusted to their tasks. In our business, we offer a clear instant value: a person should immediately see the benefits of the product.
How Technologies Penetrate the Labor Market
Daria: The behavior of people in the labor market should correspond to the organic developments in the industry: they need to constantly gain skills that are becoming relevant. Everyone has moved from SQL to business intelligence tools, Power BI and Tableau. So, people need to study these products. Everyone started doing something in ChatGPT. So, people need to take a course and learn how to write scripts for AI correctly in order to use it as effectively as possible in their work. I do not think that the emergence of new tools will make one’s expertise less relevant. You cannot become a good analyst only by mastering AI tools. But their use with an available database is a guarantee that you will be able to show high-quality results. Data analysts will not lose their jobs because of AI development, rather the opposite: the questions that they can now answer have become more complex and interesting.
Stanislav: I am an optimist and I think that if you were an expert in some field and now your job is done with the help of AI, most likely you will benefit from it. Now that you have a new tool, your attention will be directed to something else. After the advent of the AlphaFold model, which predicts protein structure, structural biologists did not find themselves unemployed. They redirected their interest and changed research subjects.
What Kind of Intermediaries Will Be Needed in the New World
Ivan: Firms do not understand the capability of AI, so there is a growing demand for confidence on their part. Confident consultants are the first to benefit. They come and say: you need to work this way, because five firms have already done it, they have earned certain revenue and you can get the same result.
Stanislav: At the first stage, you need a specialist who understands how data should be organized: if you haven't worked with it yet, most likely you just have a mess there. Therefore, people who can arrange the available data and organize it into a database will definitely be in demand. These specialists will remain in demand even after technology integration, since new infrastructure will need to be maintained. We have infrastructure laboratories that carry out standard biological measurements, and in the same way there will be departments whose employees will help you organize data and provide access to it.
Daria: In addition to organizing databases, the world of analytics is engaged with translating business issues into the language of data queries. It is important for firms to understand which results that they have obtained from their data can be used for business decisions, and which are just additional "noise". Consultants tell firms what to do with their data, and what can be done after analyzing it. And intermediaries are still needed to build a bridge from data analytics to revenue growth.
Why Fundamental Education Does Not Lose its Relevance
Stanislav: The world is changing much faster than before, so we need to learn flexibility. It's no longer advisable to rely on a clear career path in a specific field – it's more appropriate to focus on specific skills that you can use in the future in another professional sphere or with existing technologies. Classical education remains relevant in such conditions – thanks not only to knowledge, but also to soft skills that are developed during the learning process. Tools may change, but fundamental knowledge remains a valuable asset.
Ivan: Once, all education was fundamental, then its focus shifted to teaching tools; for example, learning Python and becoming a programmer. Now education is taking a step back – there is an understanding that tools can change, and code can be written with virtual assistants (like Copilot), while fundamental skills such as critical thinking, flexibility, and the ability to learn will be useful and supportive assets in any situation. Quality education should reach more people than it currently does.
Stanislav: The discussion about updating education can be divided into two parts. First, we need to offer more courses on how AI works, and open them not only in computer science faculties but also in biology and chemistry departments. Now, regarding AI, we can already talk without plunging into mathematics, explaining the essence of models, their logic, limitations, and advantages. Second, the conversation may involve restructuring education. Universities should answer the question, ”Should students be allowed to use, let’s say, ChatGPT when preparing their thesis?” Another option is to completely abandon such a format of testing final knowledge as a thesis, and some faculties are already considering this direction.
Daria: Personally, the most valuable thing in education for me was that it taught me to think and solve complex problems very quickly. Speaking about mastering new technologies, there are many ways to learn data analytics at a basic level – to master Power BI or learn programming in SQL. However, businesses need people who can think and apply knowledge in practice. That's why I always advocate for fundamental, scientific education, and after you get it, everything else will fall into place.
