Each time your digital map application calculates the fastest route to your destination, it relies on a foundational algorithm that Dutch mathematician and computer scientist Edsger W. Dijkstra invented decades ago. As a towering figure in the history of computer science, Dijkstra laid much of the structural groundwork for modern software design. Yet beyond his technical contributions to computing, he made an equally profound impact on another domain entirely: our understanding of human cognition and the very nature of thinking itself.
One might naturally assume that one of the founding architects of modern computing would have relied heavily on machines to solve the hardest intellectual problems of his day. In reality, the approach of pioneers like Dijkstra was precisely the opposite. He famously insisted on doing the vast majority of his deep, conceptual thinking entirely away from a keyboard or a terminal. Much of his most celebrated work started out as meticulously handwritten notes crafted with a fountain pen, long before any code was ever compiled or executed.
The perennial question of how we think about thinking—and, more urgently, which parts of our cognitive labor should be delegated to machines—has grown increasingly pressing in the modern era. Artificial intelligence is rapidly approaching a point where individuals and enterprises can completely offload the substance of their daily work. We are now well past the historical era of merely outsourcing physical labor to machines; today, we are increasingly capable of letting advanced AI handle the lion’s share of our creative, strategic, and analytical responsibilities.
On the surface, this technological leap appears to be an unmitigated boon, promising unprecedented levels of productivity and efficiency. After all, getting precisely what we need with minimal personal effort feels like the ultimate achievement of modern progress. However, when we outsource a task wholesale, seamlessly bypassing the internal struggle required to reach the output, we risk forfeiting some of the most valuable benefits of the work itself.
Whenever humans engage in genuine labor, we inherently produce two distinct kinds of output: the tangible work product itself and the deep learning that occurs naturally throughout the process. This internal accumulation of knowledge is what helps us solve similar, complex issues much faster the next time around. It trains our minds to recognize novel situations where we can successfully deploy our newly acquired skills, adapting past lessons to future challenges.
If society moves toward a model of problem-solving where tasks are outsourced entirely to automated systems, we will certainly continue to receive the final work product. Yet we simultaneously miss out on building the intricate internal mental models required to apply those lessons later in life. Without that foundational struggle, it becomes significantly harder to integrate new information with what we already know, leaving our cognitive architecture increasingly fragile and dependent on external aids.
The Work We Don’t Notice
One of the most insidious temptations introduced by any new technological tool is the comforting assumption that everything the tool automates was essentially wasted effort in the first place. This mindset often leads to questions of immediate utilitarian calculus: Why waste precious mental energy memorizing facts or working through equations when your phone or computer can do it for you in a fraction of a second?
Cognitive offloading is undeniably useful in a wide variety of daily situations, acting as a crucial mechanism for reducing the sheer volume of information we are forced to hold active in our heads at any given moment. Jotting down a quick shopping list on a scrap of paper or entering a medical appointment into a digital calendar frees up valuable working memory for other pressing tasks. But as technology has advanced, it has become increasingly tricky to discern which parts of a complex task carry greater long-term utility than others.
In 1998, philosophers Andy Clark and David Chalmers published a landmark paper titled "The Extended Mind," in which they argued that the traditional boundary separating human thought from the outside world is not always located where we intuitively assume it is. To illustrate this profound shift in perspective, they introduced the famous philosophical example of Otto, a man suffering from Alzheimer’s disease who relies entirely on a physical notebook to store vital information that he cannot reliably retain in his biological memory.
Because Otto consults his notebook in precisely the same functional way that another person relies on internal recall, Clark and Chalmers argued that it makes little conceptual sense to treat the notebook as merely an external aid. For all practical, operational purposes, a crucial part of Otto’s functional mind is the notebook itself.
The underlying premise is that the human mind naturally forms a deeply integrated, coupled system with its immediate environment. Our biological systems are fundamentally optimized for efficiency. If an external tool can reliably hold information or solve a specific problem on our behalf, our internal cognitive systems can afford to stop processing that information entirely. While this is a brilliant and adaptive feature of human intelligence, it also reveals why our modern choice of tools carries such high stakes. Adopting a new technology profoundly shapes the kinds of cognitive strategies we rely on, as well as the skills we ultimately allow to atrophy and die.
Paying Attention
One of the primary reasons why thinkers like Dijkstra chose to avoid computers during the deep thinking phase of their work is that writing code or prose on a keyboard is psychologically and structurally very different from writing by hand on paper.
A computer keyboard introduces a digital environment that dramatically and artificially lowers the immediate cost of making an error. A writer or programmer can type out a complex sentence, realize halfway through that it does not quite work, and instantly backspace it into total oblivion without leaving a trace. At first glance, this frictionless capability seems purely beneficial, as it famously liberates creators from the paralyzing anxiety of trying to get things perfectly right on the first try. But it also subtly and permanently alters how we approach complex problems.
When the real-world cost of making a mistake drops all the way down to zero, creators no longer feel compelled to meticulously plan their overarching structure in advance. Instead, human behavior shifts subtly away from intentional, rigorous design and toward reactive tweaking. We begin throwing half-formed, unexamined ideas onto a screen and letting those superficial prompts guide our subsequent steps.
This behavioral shift directly intersects with the concept of "desirable difficulties," a term coined by distinguished cognitive psychologist Robert A. Bjork. His extensive research and theory suggest that introducing specific kinds of friction into a learning or working environment actually improves long-term retention and comprehension. For instance, when individuals are physically forced to slow down while taking notes by hand, they lack the physical speed required to transcribe a lecture verbatim. They are forced to synthesize, summarize, and be highly selective, which inherently triggers deeper conceptual understanding. When a laptop or automated transcription tool removes that physical friction, users frequently default to passive, mindless transcription.
Arriving at Your Destination
When we systematically offload the friction of figuring things out for ourselves, we may very well accelerate our journey to a finished product. However, we ultimately arrive at our destination merely as tourists. Because we relied on a convenient shortcut to bypass the intellectual heavy lifting, we lack a deep, structural understanding of what we have just created.
If an individual only occasionally arrives at a destination as a tourist, there is generally no real harm done to their overarching intellectual development. However, if modern society moves toward a state where we constantly rely on external tools to generate the very substance of our work while we merely supervise and tweak the edges, the consequences change drastically. We will eventually end up with a polished, professional output that contains none of our actual understanding or intellectual DNA.
Decades ago, while actively contributing to the Manhattan Project during the Second World War, renowned physicist Richard Feynman observed a striking behavioral phenomenon that he memorably dubbed "computer disease." He watched with growing concern as brilliant scientists and mathematicians became so completely mesmerized by what the era’s new punch-card tabulators could mechanically accomplish that they abandoned their actual physics research to play with the machinery. They had fundamentally mistaken the act of operating a sophisticated tool for genuine comprehension of the underlying problem.
Feynman understood the vital distinction between tooling and true understanding better than almost anyone else of his generation. Throughout his remarkable career, he championed the rigorous pursuit of foundational knowledge. In a testament to this philosophy, at the time of his death, he left a single, powerful sentence scrawled across his blackboard: "What I cannot create, I do not understand."