Humans Struggle to Keep Up with AI Advancements
Norbert Wiener, the pioneer of cybernetics, famously stated, “The thought of every age is reflected in its technique.” Over the last century, our thoughts have been mirrored in our computers, particularly in the realm of AI. Google’s Demis Hassabis describes the brain as “a biological approximation to a Turing machine,” while Elon Musk bluntly refers to it as a “biological computer.” (Musk’s perspective can be a bit concerning.)
However, humans are far more intricate than a simple computer analogy suggests, and this complexity is often overlooked by the AI industry. As AI technologies permeate our culture, it becomes clear why they sometimes create chaos. AI resembles a cognitive version of a hot dog: it may be tempting in the moment, but it jeopardizes the health and sustainability of the cognitive systems we’ve developed over millennia.
To grasp the issue, we need to delve into some neuroscience and the entirety of human evolutionary history.
The notion that our brains function like computers dates back to Alan Turing, who proposed that thought operates as a three-step process: our minds receive information (input), process it (computation), and then produce behavior (output). This leads to a sequence where perception informs cognition, which then drives action. In more detailed terms, it can be visualized in a flowchart.
This model has been useful in many respects. By conceptualizing our brains as computers, technologists have advanced computing from basic calculators to sophisticated neural networks and generative AI, striving to create ever more accurate reflections of our minds.
Yet, this reflection is not entirely accurate; it’s distorted and limited. John von Neumann, a key figure in computer science, expressed skepticism about whether the computational model could truly capture the “exceptional complexity of the human nervous system.” Our nervous systems evolved to help us navigate the vast ecological landscape we call “the world,” allowing us to exert control over it as best we can.
The computational approach focuses on the end results of our mental processes and attempts to “reverse engineer” their functions. Instead, we might benefit from examining the long arc of evolutionary history to construct our model from the ground up. This is the perspective of Paul Cisek, a neuroscientist at the University of Montreal, who has developed a biological model of brain and nervous system evolution over millions of years.
Cisek argues that our brains should be viewed not merely as information processors but as feedback-control systems. Our bodies don’t just receive input; they actively adjust that input based on available options. Cisek points out that this idea isn’t new — philosopher John Dewey described the mind as a circuit, where the motor response influences the stimulus just as sensory input drives movement.
What distinguishes these two approaches?
Consider a classic baseball scenario: catching a fly ball in the outfield. The computational model suggests that this requires complex, subconscious calculations where the outfielder estimates the ball’s speed, factors in gravity, and performs numerous mathematical operations to determine where the ball will land.
In contrast, the feedback-control model proposes a straightforward heuristic: “keep the ball in the same position in your visual field and move to maintain that position.” This approach aligns more closely with the experiences of anyone who has played center field, avoiding the need for intricate calculations that the computational model demands. Humans operate in a more dynamic manner.
This feedback-control model also aligns well with the biological evolution of our brains. From ancient fish to mammals and eventually modern humans, we see that new behaviors arise in response to changing environmental conditions. For instance, when dinosaurs went extinct, some nocturnal creatures could venture out during the day with less risk, leading to the development of new capabilities. Cisek notes that the history of our nervous system is one of “continuous extension of control further and further into the world.” (His upcoming book will delve deeper into this topic, and I hope this essay nudges him to finish it.)
This leads to a different diagram than the one previously mentioned. We can map behaviors and abilities as they emerged over time and overlay the specific physical components of the brain associated with these capabilities.
Instead of progressing from left to right as in the computational model, this model unfolds from top to bottom over evolutionary time. For example, as vertebrates evolved greater mobility, specialized systems for exploration became advantageous, leading to the development of structures like the hippocampus. This also enabled our ancestors to remember significant moments, giving rise to “episodic memory.”
The computational model fails to account for this complexity; it doesn’t align with observable neuroscientific structures and obscures the fact that much of what our brains do involves managing interactions with our environments. Cisek argues for a significant shift in how we understand the relationship between our brains and behavior, moving away from algorithmic input-output models toward more dynamic feedback systems.
So far, our exploration of human development has focused on feedback from the physical world. However, one of humanity’s most crucial “feedback loops” stems from our inherently social nature. The computational model, along with the AI industry, struggles to address this aspect. Over thousands of years, we’ve built institutions and norms for learning and communication, yet in less than a decade, major tech companies have systematically undermined them.
To understand this, we need some historical context.
Hundreds of thousands of years ago, our ancestors achieved something remarkable: they learned to imitate one another. By mimicking gestures and movements, they could share successful practices, like crafting stone tools, and coordinate complex activities through shared rituals, which in turn shaped their cognition. This marked the dawn of human culture.
Humans progressed to imitating sounds, leading to the development of oral language. While language is not synonymous with thought, it allows us to express our ideas. Eventually, our nomadic hunter-gatherer ancestors settled into communities, developing agriculture to cultivate food rather than constantly migrating.
We then began using written symbols to represent spoken language and abstract concepts, enabling the transmission of complex ideas across generations. Formal education emerged to ensure knowledge was shared within communities, along with institutions that facilitated collective decision-making, such as markets, laws, and democracy.
Each of these claims is debatable and deserves more detail than can be provided here. However, this process is intertwined with our biological evolution, as both foster a broader range of control over our surroundings.
Yet, cultural changes can lead to unintended consequences that may be detrimental.
Human diets illustrate this well. In our hunter-gatherer days, fatty foods were scarce but vital for survival — just watch any survival show to see this in action — and we evolved to seek and store them. However, the advent of agriculture has made fatty foods abundant today — if you want a hot dog, they’re everywhere.
This is the crux of the issue. Hot dogs can harm our physical health by clogging arteries with unnecessary fat. Similarly, AI threatens our cognitive abilities by hindering our capacity to develop essential knowledge for navigating the world. It’s a cognitive hot dog.
An occasional hot dog won’t hurt anyone, but it becomes problematic if it becomes a staple of our diet. The same applies to AI — the danger lies not in its occasional use but in integrating it into our mental routines.
Yet, that’s precisely what major tech companies are pushing for.
In a July 2025 podcast discussing “how AI is transforming education,” a tech executive claimed that “learners” comprised over half of the 900 million average monthly users of a popular AI tool, which was touted as “the world’s largest learning platform.” Recently, another company announced a new initiative providing free access to premium AI models for teachers.
These initiatives won’t benefit students or educators. They exemplify how AI undermines the systems we rely on for cognitive development in the name of convenience.
A former software engineer has described AI as a tool for “cognitive automation,” which involves encoding human abstractions into software to automate tasks typically performed by humans. An interdisciplinary group of scholars has argued that we should view generative AI not merely as intelligent agents but as a new kind of cultural and social technology that allows humans to leverage accumulated knowledge.
AI advocates have made numerous predictions about its benefits, often dismissing the drawbacks as comparable to earlier forms of automation, like computers and calculators. (I’ve encountered many references to Plato’s supposed opposition to writing, which I’ve addressed separately — he was a writer!)
Regardless of the impact of earlier technologies, we’ve never before developed something explicitly designed to replace human thinking — a point emphasized by AI proponents themselves.
Take education, for example. While a popular AI tool may be the largest learning platform, students are increasingly using it to bypass the hard work necessary for genuine understanding. Evidence is mounting regarding the negative effects of AI tools in educational settings (and many students dislike it). A recent study from China found that thousands of students stopped doing their homework after starting to use AI, which significantly hindered their learning. There’s even evidence that using large language models for learning purposes can lead students to relax their judgment and critical thinking in other contexts.
No other species has a brain and body that develop over such an extended period as ours, and historically, we’ve complemented this with unique cultural institutions that transmit knowledge across generations. We should fiercely protect this distinctly human endeavor, yet many educational administrators and tech companies are aggressively pushing AI into our educational systems.
If we view the mind solely as a computer, AI might seem beneficial for learning — which is why some envision AI tutors. However, such efforts are already faltering because our minds are more than just input-output machines. We exist within a broader cognitive ecosystem where we must make choices about how to allocate our cognitive energy. We do this to enhance our agency and control over our environment, however limited that may be.
This presents a challenge. Engaging in effortful thinking is difficult, and when offered technological shortcuts, we’re likely to take them — a phenomenon some researchers have termed cognitive delegation. As cognitive delegation becomes more prevalent, they warn, “the social environment that supports autonomous reasoning can weaken,” making such delegation even more appealing.
The researchers advocate for a program of cognitive immunization against AI, one that preserves “the practices and institutions that keep human cognition active: unaided problem solving, verification, critical discussion, and periods of deliberate disengagement” from AI.
We refer to these institutions as “schools.”
Policy changes are also necessary. It’s noteworthy that Norway recently banned nearly all uses of AI in schools for students under 13. In the U.S., a major teachers’ union has called for prohibiting chatbots in elementary schools, and significant school districts like Los Angeles and New York City have implemented similar bans. (I’ve informally advised an organization that advocated for these bans.) They likely won’t be the last to do so.
The silver lining, if it can be called that, is that resistance is growing — perhaps you’ve seen videos of college graduates booing speakers who praise AI? This backlash is likely fueled by job concerns, but many recognize the harm AI has done to their cognitive development. This sentiment is echoed by students at the Oberlin Luddite Club, who wrote an open letter to their university president criticizing the institution’s embrace of AI: “As you embark on your year of AI, we’ll embark on our own year of self-actualization — of realizing the fruits of our labor, and of embracing human imperfection and raw inquiry.” Right on.
The control systems in our brains are the result of millions of years of biological evolution and thousands of years of cultural evolution. We need to move beyond the computational model of the mind to fully understand what makes us human and to appreciate our deeply social nature, which drives human culture. We’ve shown remarkable resilience as a species, but indulging in cognitive hot dogs isn’t beneficial for any of us.
Eat healthy, and think healthy.
