In Russia, Machines Learn to ‘Think’: Yandex Searches for Traces of Consciousness in AI
Russian researchers are running large-scale experiments to determine whether neural networks can form their own viewpoints rather than simply echoing narratives written by humans.
Scientists in Russia are conducting an unprecedented research project to explore whether artificial intelligence is capable of developing stable preferences—something akin to a personal viewpoint—rather than reproducing patterns extracted from human-produced texts.
The study is being carried out by Yandex in collaboration with Konstantin Anokhin, a neurobiologist and Full Member of the Russian Academy of Sciences. Their joint goal is to answer a fundamental question: can artificial intelligence truly ‘think,’ rather than merely process data?
A Neural Network’s ‘Point of View’
According to the research team, the core of the experiment focuses on identifying whether an AI system can exhibit internal consistency — meaning the ability to maintain a position or preference even when confronted with conflicting data. In other words, can a machine defend its own stance rather than rely on a collage of learned textual patterns?
Detecting early signs of what could be interpreted as ‘consciousness’ would open new avenues for AI development. Such systems could move beyond surface-level responses, offering original arguments or independent evaluations of situations.
The Nature of Consciousness
The project may also illuminate broader questions about the origin of consciousness — not only in humans but in digital systems as well. However, key uncertainties remain unresolved: what qualifies as ‘consciousness’ in the context of AI, and which behavioral or computational markers would justify claiming that an artificial system has its own viewpoint?
The work sits at the intersection of neurobiology, cognitive science, and advanced AI modeling, enabling researchers to examine the limits of what neural networks can achieve.
The Russian initiative highlights a significant turn in AI research: technological companies and scientists are stepping beyond engineering tasks and addressing deep philosophical questions about machine cognition, the possibility of non-human reasoning, and the long-term prospect of AI becoming an intellectual partner to humans.
