Google DeepMind lead researcher Alexander Lerchner reached an unequivocal conclusion in his recent scientific paper: artificial intelligence, like any other computer system, will never be able to become conscious.
This statement is particularly telling against the backdrop of technology industry leaders actively promoting the idea that artificial general intelligence (general-purpose AI), which could supposedly rival human consciousness, will soon emerge.
Recently, large language models have demonstrated such impressive results that the question of whether they might develop some form of consciousness has increasingly arisen both in scientific circles and in public debate. The rapid growth in computing power suggested that, over time, AI might become complex enough to develop sophisticated behavior traditionally regarded as a uniquely human trait.
Back in 2022, Ilya Sutskever himself, co-founder and former chief scientist of OpenAI, suggested that “today’s large neural networks may be slightly conscious.” Executives at companies developing AI are actively using this hypothetical possibility to attract investment in the concept of general-purpose AI, which promises a global socioeconomic revolution. Moreover, the very idea of conscious AI has sparked serious debate about welfare and moral responsibility in relation to such technologies. Nevertheless, the conscious AI hypothesis remains highly controversial.
A number of studies have already indicated that mathematical algorithms implemented on graphics processing units are incapable of developing consciousness because they lack a complex biological substrate. However, their advanced linguistic capabilities, designed to imitate human speech accurately, can easily create a misleading impression of their true nature. It is precisely this position that Alexander Lerchner of Google DeepMind reinforces in his work.
In his paper, published on the company’s internal server, he writes that functionalism, which dominates contemporary debates about AI consciousness, is given a physically grounded refutation that can resolve the current uncertainty surrounding this issue.
Functionalism is the hypothesis that the subjective experience underlying consciousness can arise from an abstract causal topology regardless of the specific physical substrate.
Lerchner firmly rejects this approach, arguing that any AI system is inherently dependent on an active cognitive agent—a human developer who receives and organizes sequences of ideas and data. On its own, such a system is incapable of developing complex, consciousness-like behavior. Put simply, the researcher insists that consciousness cannot exist without a complex physical body, a living organism.
Mark Bishop, Professor of Cognitive Computing at the University of London, commented on Lerchner’s conclusions by saying that he agreed with 99 percent of his ideas, while noting that all such arguments had already been presented many years ago. Biologist and philosopher Johannes Jaeger, in turn, was more critical, suggesting that the researcher had arrived at these conclusions independently, had “reinvented the wheel,” and displayed insufficient familiarity with philosophical issues and biology.
Jaeger attributed this to the relative insularity of the AI research community, which focuses on narrow fields and works under intense pressure without sufficient time to study related disciplines. Both experts agreed, however, that the very fact that a Google DeepMind employee had published such a study was positive, although it seemed surprising that his employer had approved it.
Mark Bishop suggested that Google may have compelling financial and legal reasons to support the conclusion that computation is incapable of producing consciousness. According to him, if the opposite were true, attempts would already be under way in Europe to pass legislation granting rights to computer systems, which seems absurd.
Accordingly, Google would be very pleased if its systems were not considered conscious, as this reduces the risk of additional regulation in the United States and other countries around the world. At the same time, Lerchner’s study itself includes a disclaimer stating that its conclusions may not reflect his employer’s official position, views, or strategic policy.
Source: https://new-science.ru/issledovatel-google-deepmind-dokazal-chto-iskusstvennyj-intellekt-nikogda-ne-obretet-soznanie/
