...

So, at past conferences, we talked in detail about generative artificial intelligence, about language models trained on a huge array of data, including texts and images. As recently as two years ago, such systems could perform rather simple tasks, and today, on the basis of generative artificial intelligence, intelligent assistants, so-called AI agents, as well as unmanned transport and robots are already being created.

What is important, they do not act according to rigid, preset algorithms, but make increasingly autonomous decisions without human participation. Those who already own and use such products gain significant advantages in efficiency and productivity.

It is clear that generative artificial intelligence technologies are becoming crucial and strategic. Not only the largest companies, but also leading countries, are competing for their own fundamental language models.

According to various estimates, the development of artificial intelligence is perhaps one of the largest technological projects in history. At the same time, the main focus of investment is on increasing computing power and generating additional gigawatts of energy, which is crucial for the reliable operation of the growing "digital infrastructure." This is necessary for training the next generation of more efficient generative artificial intelligence systems. As I mentioned earlier, data center electricity consumption is expected to triple over the next decade.

Along with rapid changes in all areas, the number of users of products based on generative artificial intelligence is also increasing. In Russia, the vast majority of young people are already using neural networks and language models extensively for various tasks in their studies, careers, and daily lives.

But another important thing is that such models generate gigantic amounts of new data and become one of the most important tools for disseminating information, which means that they can influence people's values and worldviews and shape the semantic space of entire countries and even humanity as a whole.

We cannot allow ourselves to become critically dependent on foreign systems. For Russia, this is a matter of state, technological, and, I would say, value sovereignty, which is why our country must have a comprehensive set of its own technologies and products in the field of generative artificial intelligence.

First of all, these are national language models, both fundamental and small, for specific industries. The entire range of such models should be trained and fully controlled by Russian specialists at all stages, including the verification of the final product.

Working on such complex systems is intended to become a serious incentive for strengthening the domestic engineering school in this most important area. We must have unique competencies in the full cycle of creating fundamental language models: from their creation and training to their adaptation to the needs of various industries. In fact, this is the path taken by all countries that seek to ensure their leadership and independence, their sovereignty in this area.

...