Is AI Actually 'Thinking'? Revisiting a 1984 Soviet Critique
The 40-Year-Old Debate on Machine Consciousness
It is easy to get caught up in the current hype cycle of generative AI. With new models launching seemingly every week, we often forget that the philosophical questions we are asking today—'Is this thing actually thinking?' or 'Is it just a stochastic parrot?'—have been echoing through history for decades.
One of the most fascinating artifacts from this long-standing debate comes from an unexpected source: the Soviet Union in the 1980s. In 1984, a book titled Cybernetics Today: Achievements, Challenges and Prospects, published by Mir Publishers in Moscow, featured a chapter by Soviet psychologist A. V. Brushlinsky titled 'Why Artificial Intelligence Is Impossible' [1].
While the title sounds like a relic of a bygone era, the arguments within it are eerily similar to the debates we see on platforms like Reddit and Hacker News today [3]. Let’s peel back the layers of this historical critique and see what it tells us about the nature of intelligence.
The 'Disjunctive' vs. The 'Holistic'
To understand Brushlinsky’s skepticism, we have to look at his core philosophical framework. Brushlinsky argued that machines are fundamentally 'disjunctive' [1]. By this, he meant that computers are composed of separate, discrete, and algorithmic parts. They operate on a sequence of logical steps, which he believed could never replicate the 'holistic' nature of human mental processes [1][2].
In his view, human consciousness is an integrated whole—it is not merely the sum of its logical parts. Because machines rely on discrete operations, he posited that they could simulate the results of human thought, but they could never truly possess the process of thinking itself [2].
This perspective reflects a specific Soviet philosophical tradition known as dialectical materialism, which viewed psychology and consciousness as complex biological phenomena [2]. To these thinkers, intelligence was an emergent property of life, not something that could be reduced to mathematical operations or code [2].
Why It Still Matters Today
If you have followed the discourse around Large Language Models (LLMs), Brushlinsky’s argument might sound familiar. Modern observers often note that his critique bears a striking resemblance to John Searle’s famous 'Chinese Room' thought experiment, proposed in 1980 [2].
Just as Searle argued that a machine manipulating symbols does not 'understand' the language it processes, Brushlinsky was warning us about the gap between simulation and genuine understanding [2]. Today, when we debate whether an AI 'knows' what it is saying or is simply predicting the next token in a sequence, we are essentially re-litigating the same philosophical battle that Brushlinsky fought in the 1980s [3].
It is a recurring irony: while the technological capabilities of 2026 are lightyears ahead of 1984, the fundamental question of what it means to 'think' remains just as elusive.
A Nuanced View of Soviet Science
One common misconception is that this text represents a unified, state-mandated doctrine of the USSR. That is not the case. Cybernetics Today was an anthology that included a wide variety of viewpoints, ranging from ardent proponents of AI development to vocal critics [1].
This diversity of thought highlights that the Soviet scientific community was not a monolith. There were deep internal disagreements about the feasibility and ethics of cybernetics and AI, much like the fragmented opinions we see in the global tech community today [3].
Frequently Asked Questions
How representative was Brushlinsky's view of the broader Soviet scientific community?
Brushlinsky’s view was one of several competing perspectives within the Soviet Union. The book Cybernetics Today served as a platform for diverse voices, meaning his skepticism was part of a broader, ongoing academic debate rather than a single, enforced state doctrine [1].
Did this specific paper influence Soviet AI funding or research direction?
There is no clear evidence that this specific chapter served as a direct policy lever to halt funding. Instead, it serves as a historical record of the intellectual climate at the time, reflecting the philosophical tensions that existed alongside the practical efforts to advance cybernetics in the USSR [1].
As we continue to push the boundaries of Artificial General Intelligence (AGI), perhaps the lesson from 1984 is not that AI is impossible, but that our definition of 'intelligence' needs to be as dynamic as the technology itself. Does Brushlinsky’s 'disjunctive' argument hold up against the massive, interconnected architectures of today’s neural networks? That is a question for us to answer.
