Russian Neuroscientists Aim to Teach AI to “Think” and “Create”
Researchers from Moscow State University and Yandex are working on a brain-inspired AI system that moves beyond statistical prediction, with the goal of giving machines more human-like reasoning and creativity.
Russian scientists and engineers are launching an ambitious experiment at the intersection of neuroscience and artificial intelligence. Neurobiologists from Lomonosov Moscow State University and developers from Yandex are collaborating on a neural network designed to think and create “like a human,” according to the project’s researchers.
The effort brings together insights into how the human brain works with modern AI development. The team is studying how biological neural networks learn, store information, and retain skills over a lifetime—and how those mechanisms might be translated into artificial systems.
Learning From the Brain
One of the key figures involved is Konstantin Anokhin, director of MSU’s Institute for Advanced Brain Studies and a leading Russian expert in neuroscience. In the experiments, researchers observe how biological neural networks acquire knowledge and preserve memory across their entire lifespan—capabilities that remain a major challenge for artificial neural networks.
A Microscope for Neural Networks
As part of the project, researchers at MSU are assembling a two-photon laser microscope. The system makes it possible to observe neural connections with a level of precision and speed unattainable using classical research methods. Data collected through these observations is expected to form the basis for developing a digital model of intelligence that more closely mirrors the brain’s inner workings.
Moving Beyond “Average” AI
At Yandex, developers say the collaboration is aimed at moving past what they describe as a “statistical average” approach to AI. Today’s neural networks generate responses largely by predicting the most likely answer based on data patterns, without forming genuine preferences or internal logic.
The researchers want to test whether an AI system can develop its own internal reasoning if it is trained not only on external data, but also on its own previous conclusions and responses. In theory, that feedback loop could push neural networks closer to true reasoning—and eventually, creative problem-solving.
Probing Consciousness, Not Just Code
The experiment is framed as more than a technical exercise. Scientists involved say it could also shed new light on the mechanisms of human consciousness itself. As a point of comparison, researchers are examining how today’s AI systems “think.” One student at Bauman Moscow State Technical University described asking AI models to solve problems step by step, not just to see the final answer, but to analyze the reasoning process behind it.
Russia is already among the countries actively deploying AI technologies across the economy. The domestic AI market continues to grow, and developers say the first results of Yandex’s brain-inspired neural network experiment are expected to be published openly in the near future.
If successful, the project could challenge a core assumption of modern AI—that intelligence must be statistical—and replace it with something closer to how the human mind actually works.
