AI Is Junk Food for the Mind

Overview

For decades, the dominant metaphor in technology has been that the human brain is essentially a biological computer. This computational model suggests a linear flow: input arrives, the mind processes it through an algorithmic mechanism, and output emerges as action. Prominent figures in the AI industry have amplified this view, with Google DeepMind’s Demis Hassabis describing the brain as a "biological approximation to a Turing machine" and Elon Musk bluntly stating that people should simply think of the brain as a biological computer.

However, this analogy is increasingly viewed by neuroscientists and philosophers as dangerously reductive. As artificial intelligence products embed themselves deeper into culture, critics argue that this mismatch in understanding explains why these tools often undermine human cognition. The issue is not merely that AI is inaccurate, but that it operates on a flawed premise of how human thought actually functions.

"The thought of every age is reflected in its technique." — Norbert Wiener

Norbert Wiener, the godfather of cybernetics, famously noted that our techniques mirror our thinking. For the past century, we have been building machines that reflect a simplistic, input-output view of the mind. But humans are far more complex than this model allows. Treating AI as a cognitive shortcut is akin to consuming junk food: it is immediately tempting and easy to digest, but it ultimately erodes the long-term health of the systems we evolved to sustain ourselves.

The Computational Model and Its Limits

The idea that brains are computers traces back to the foundational work of Alan Turing. This framework posits that thought is a three-step process: our minds take in information from the world, manipulate it internally, and then enact behavior. In this view, perception leads to cognition, which in turn leads to action. It is a clean, logical structure that has driven the rapid advancement of computing from simple adding machines to complex generative neural networks.

Yet, this mirror held up to the human mind is warped. John von Neumann, a seminal figure in computer science, doubted that the computational model could ever capture the "exceptional complexity of the human nervous system." The core flaw lies in the fact that our nervous systems did not evolve in a vacuum to process abstract data; they evolved to help organisms navigate the vast, ecological system we call the world.

The computational approach attempts to "reverse engineer" the mind by looking at the end product of our thoughts. It ignores the context of survival, physical interaction, and evolutionary history. By focusing solely on the output, it misses the dynamic feedback loops that define conscious experience. Humans are not passive recipients of data waiting for a processing cycle; we are active agents constantly shaping our environment.

The Feedback-Control Alternative

A more accurate understanding of the brain comes from viewing it as a feedback-control system rather than an information processor. This perspective is championed by Paul Cisek, a neuroscientist at the University of Montreal, who argues that brains are better understood through millions of years of biological development. In this model, the body does not just receive input; it takes action to adjust what that input is, based on available options.

This concept is not entirely new. At the turn of the 20th century, philosopher John Dewey described the mind as a circuit that is "more truly termed organic than reflex." He argued that the motor response determines the stimulus just as truly as sensory stimulus determines movement. We act upon the world to exercise control over it, and our perceptions are shaped by those actions.

Consider the classic example of a baseball outfielder catching a fly ball. The computational model would suggest the brain performs complex subconscious calculus: estimating velocity, calculating gravity’s effect, and solving differential equations to predict the ball’s trajectory. In reality, the outfielder uses a simple heuristic: keep the ball in the same position within their visual field and move to maintain that situation. This approach avoids separating mental processes from physical movement and relies on dynamic adjustment rather than static computation.

Evolutionary History and Cognitive Health

This feedback-control model aligns neatly with the biological architecture of brains as they have evolved. From ancient fish to amphibians, mammals, primates, and eventually modern humans, new behaviors emerged in response to new environmental possibilities. When dinosaurs died out, for instance, nocturnal creatures could roam safely during the day, giving rise to new navigational capacities.

Cisek observes that the history of our nervous system is one of "continuous extension of control further and further into the world." This includes the development of specialized systems for exploration, such as the hippocampus, which evolved to help vertebrates navigate using landmarks. Instead of a left-to-right pipeline of input-to-output, human cognition is better mapped as unfolding from top to bottom over evolutionary time, layering complex social and abstract reasoning onto ancient survival mechanisms.

AI, by contrast, offers a shortcut that bypasses these evolved feedback loops. It provides the "output" without the necessary cognitive "process" or physical engagement with the world. Just as junk food provides calories without nutrition, AI provides answers without the mental exertion that strengthens cognitive resilience. As these tools become ubiquitous, the risk is that we may atrophy the very skills—pattern recognition, spatial navigation, and adaptive problem-solving—that have sustained human intelligence for millennia.

To harness AI responsibly, we must move beyond the crude metaphor of the brain as a computer. Recognizing our minds as dynamic, feedback-driven systems allows us to design tools that augment rather than replace our natural cognitive processes. Only by respecting the evolutionary complexity of human thought can we ensure that AI serves as a tool for enhancement, not a substitute for understanding.