In a world where neural networks write poetry and generate art, the question of whether they possess consciousness has ceased to be the sole domain of science fiction. However, a new study by an international team of scientists led by the Institute for Basic Science (IBS) calls for caution. Experts argue: we may simply be mistaken, attributing human feelings to machines where there is only complex mathematics.
The Trap of Cognitive Tests
The problem lies in the measurement tools themselves. Modern methods for testing consciousness — from sensory threshold tests to visual masking checks — were originally developed for biological organisms. Scientists from IBS, the University of Montreal, and New York University have found that these tests measure only the system's ability to process information, not the presence of subjective experience.
As Hakwan Lau, Director of the Center for Neuroimaging Research, notes, developers inadvertently fall into the trap of anthropomorphism. When AI demonstrates a reaction to a stimulus, we tend to consider it proof of "feeling," whereas in reality, it may simply be a flawless algorithmic response.
Lessons from the Past: The Crisis of Psychology
The history of science knows precedents for such misconceptions. At the turn of the 19th and 20th centuries, psychology experienced a crisis due to loud but unfounded claims about the nature of consciousness. This led to the rise of behaviorism — a movement that, for decades, forbade researchers from mentioning the "inner world," focusing exclusively on observable behavior.
Today's researchers fear a repetition of this scenario in the field of artificial intelligence. If we continue to confuse data processing with experience, we risk either creating false ethics for machines or, conversely, missing the moment of their true awakening.
Blindsight as a Key to the Truth
To distinguish between "intelligence" and "consciousness," scientists suggest studying specific neuropsychological phenomena. A striking example is "blindsight." In this state, a person can react to objects they do not see and are not aware of. This proves that the brain is capable of processing information without the involvement of subjective perception.
Applying such models to AI will allow for the creation of stricter criteria. Only by isolating subjective experience from simple cognitive processing can we honestly answer the question: are our digital children alive, or are they just very smart calculators?