AI and the Future of Human Intelligence
If intelligence is humanity’s defining trait, then our relationship with those tools that shape thinking—cognitive artifacts—is among the most consequential forces in history. From the earliest tally marks on bones to modern artificial intelligence, humans have built external supports for cognition that both extend and transform our mental capacities. But do these artifacts make us smarter or, paradoxically, lead to cognitive decline.
Few have wrestled with this question as provocatively as David Krakauer, an evolutionary theorist and president of the Santa Fe Institute. Krakauer has argued that while cognitive artifacts can enhance human intelligence, they may also erode it.
His framework hinges on a crucial distinction between those cognitive artifacts that are “complementary” and those are “competitive”:
Complementary Cognitive Artifacts
Complementary cognitive artifacts amplify human cognition by integrating with our mental faculties, sharpening and expanding them. Writing—from a quick scribble to an accountant’s notebook and vast manuscripts of literary fiction and non-fiction, but also mathematical notation, and maps—are classic examples: external supports that make thinking more precise and, importantly, scalable.
And, ultimately, we aren’t dependent on them: once we have mastered the physical artifact, we rely on a virtual one our minds have created —think of the ABC chart on the wall above the chalkboard in your elementary school years ago, or that map you no longer need…
Competitive Cognitive Artifacts
Competitive cognitive artifacts, on the other hand, replace cognitive functions rather than support them—the way an escalator replaces that strenuous walk up a flight of stairs—and they lead, potentially, to atrophy. Calculators, GPS navigation, and predictive text fit this category; they perform cognitive tasks on our behalf, removing the need to develop or maintain certain skills ourselves.
Said another way: complementary cognitive artifacts are teachers; competitive cognitive artifacts are serfs—or, that’s until Ray Kurzweil’s Singularity.
At its core, Krakauer’s model is a cautionary tale: as we increasingly outsource our thinking to AI, automation, and digital tools, we risk hollowing out our own intelligence. If competitive cognitive artifacts were inherently dangerous, civilization itself could, I believe, never have emerged as it has.
To understand why, let’s imagine Krakauer transported back to ancient Mesopotamia—where he might find himself deeply unsettled…
The Cognitive Shock of Civilization
If Krakauer were to find himself in Uruk, he would witness the world’s first large-scale experiment in cognitive outsourcing. There, on the banks of the Tigris and Euphrates, humans had abandoned the cognitive demands of foraging—the need for deep ecological knowledge, spatial navigation, and food acquisition strategies—in favor of something unprecedented: agriculture.
By Krakauer’s assertion, this was the dawn of a competitive cognitive artifact. Farming externalized and simplified what had once been a constant, adaptive, intellectual challenge. The need to track animal migrations and remember which plants were safe to eat had been replaced by fixed fields, domesticated crops, and surplus grain storage. Surely, according to his argument, this should have degraded human cognition.
And yet, it didn’t.
The stability of agriculture freed up cognitive resources for new developments: record-keeping, formalized trade, governance, and abstract reasoning. The people of Uruk developed writing not as literature, but as bookkeeping—cuneiform imprints on wet clay to track grain and livestock. Over time, this practical system of external memory evolved into written language, setting the foundation for philosophy, law, and science.
And, of course, there is more.
Let’s consider some of the greatest leaps in human history—all of which are innovations that liberated energy, cognition, and time. These advancements arose from the biological imperative to limit the expenditure of energy, as I argue in my next book Please Make Yourself (Un)comfortable. According to a strict interpretation of Krakauer’s model, they might be classified as competitive artifacts:
Writing
Writing is, at the very least, an obvious externalization of memory that allowed humans to offload the burden of oral tradition. If Krakauer is right, writing should have weakened our ability to retain information. And yet, counterintuitively, it created a world in which humans could think more abstractly, build on past knowledge, and —over generations—engage in long-term, cumulative reasoning.
Mathematical Notation
Once upon a time, people had to hold complex numerical relationships in their heads, performing intricate mental calculations. Then came numerals, equations, and, eventually, computers. If Krakauer’s model were absolute, these developments should have dulled our mathematical minds. Instead, they have enabled quantum physics, machine learning, and space exploration.
Mechanization and the Industrial Revolution
Mechanization took over an enormous range of cognitively embedded human labor, making people far less adept at the artisanal skills that defined pre-industrial societies. But in doing so, it allowed the human mind to redirect its focus toward engineering, economics, and systems design.
These examples—from writing to mechanization and the Industrial Revolution—suggest that cognitive outsourcing with competitive cognitive artifacts doesn’t necessarily lead to decline, as Krakauer would agree if pressed; in fact, it can be an accelerant. The paradox of lowering energy expenditure and freeing up energy and time inherent in the use of competitive artifacts is that by removing certain cognitive burdens, they create the very conditions for new kinds of intelligence to emerg
So, if cognitive outsourcing inevitably weakens intelligence, how do we explain the fact that competitive cognitive artifacts have repeatedly enabled intellectual revolution
To resolve this contradiction, a sharper distinction between different kinds of cognitive artifacts is required—not just competitive versus complementary, but weakening versus enabling.
The Competitive Artifact Paradox – where the transformation occurs
To understand this paradox, consider two historical examples:
1. The Abacus vs. The Digital Calculator
The abacus is a classic complementary cognitive artifact. It doesn’t solve problems for you—it structures your thinking. By physically manipulating beads, users develop a deep intuition for arithmetic. Remarkably, expert abacus users can often visualize the tool in their minds and perform calculations without needing a physical device at all. The abacus, in this sense, trains cognition rather than replacing it.
The digital calculator, by contrast, is a competitive cognitive artifact. It delivers instant answers without requiring internalized mathematical reasoning. A person using a calculator can quickly compute 1,237 × 456, but if you take that device away, it’s possible they will struggle to reconstruct the steps. Yes, the calculator is useful, but it doesn’t develop or reinforce mathematical intuition—it replaces it. It is cognitive offloading that pays no dividends—in fact, we pay for it.
2. Navigational Memory vs. GPS
Human navigation once depended on a detailed internal map of landscapes, star positions, and environmental cues. And the distances these voyagers traveled were substantial. For example, early Polynesians settled far-flung islands across the Pacific, ranging from Hawaii in the north to Rapa Nui (Easter Island) in the east and New Zealand in the south. These voyages could span over 2,500 miles, such as the journey from the central Pacific islands to New Zealand. The navigation between Hawaii and Tahiti, another common route, is about 2,600 miles. For a modern human—or perhaps just me—this seems nearly impossible.
Today, GPS has largely replaced the need for spatial reasoning in our navigation. Studies show that reliance on GPS correlates with reduced hippocampal activity—the brain region associated with spatial memory. In extreme cases, as many can attest, people become unable to navigate without it. Unlike maps, which require active interpretation and integration, GPS offers instant directions, leading to what Krakauer would call cognitive atrophy.
But, now, here’s the twist: while both the calculator and GPS can diminish certain cognitive skills—such as mental arithmetic or wayfinding—they also enable new forms of intelligence. The calculator minimizes the drudgery of routine arithmetic, freeing up cognitive load and time for mathematicians to tackle more advanced concepts in algebra, calculus, and theoretical research. At the same time, GPS goes beyond basic navigation by empowering individuals, businesses, and governments to optimize logistics, plan urban infrastructures, and streamline global transportation networks—activities that demand new kinds of spatial reasoning and data-driven decision-making.
This suggests a crucial refinement to Krakauer’s model: not all cognitive outsourcing with competitive cognitive artifacts is intellectually destructive—and this framing of cognitive artifacts distracts from the real issue. Some of them shift our cognitive burdens in ways that free up time and, importantly, the necessary energy to allow higher-order reasoning to flourish.
A More Nuanced Model: When Cognitive Outsourcing Becomes an Upgrade
If some competitive artifacts lead to decline while others create new frontiers of intelligence, how can we tell the difference?
Consider The Civilization-Scaling Effect and The Cognitive Autonomy Factor.
1. The Civilization-Scaling Effect
Let’s take two different kinds of cognitive outsourcing:
Competitive Weakening Artifacts are those that replace cognitive effort without enabling more advanced reasoning. Think of passive content consumption, automated algorithmic decision-making that eliminates the need for critical thought, or predictive text systems that cause our writing skills to atrophy.
Competitive Enabling Artifacts are those that shift cognitive burdens in a way that frees the mind for more abstract or strategic thinking. Writing systems might weaken rote memory, but they also enable complex argumentation. Supply chains might eliminate the need for individuals to manage logistics and eliminate self-sufficiency, but they create the conditions for global economic coordination and global scientific collaboration.
Thus, the danger Krakauer warns of is real, but the true issue isn’t cognitive offloading itself—it’s how we structure and integrate our cognitive artifacts…
2. The Cognitive Autonomy Factor
Another crucial question: Does reliance on an artifact make us passive or active thinkers?
AI-Generated Text: If AI replaces writing, it could weaken composition skills. But if AI helps organize ideas, enhance synthesis, and spark new insights, it might be the tool for intellectual acceleration rather than atrophy.
Search Engines: Memorization of facts may degrade, but search engines allow for broader interdisciplinary thinking. The important question is: do we use them to expand our knowledge or as a crutch for avoiding deep engagement?
Yes, Krakauer is right to warn of cognitive atrophy, but his warning should be about how cognitive artifacts are structured and utilized rather than whether they exist at all. The key isn’t whether an artifact replaces cognition but whether it fosters a shift toward more sophisticated cognitive tasks.
And this suggests a guiding principle: the best cognitive artifacts are those that don’t just provide answers but demand engagement and synthesis.
Implications for AI, Education, and the Future of Intelligence
If we accept this revised framework, it changes how we should think about AI, education, and civilization itself.
AI as a Cognitive Accelerator
There’s a growing anxiety that AI—especially tools like ChatGPT—will make humans lazy thinkers. And it’s a legitimate concern: if AI is used as a shortcut to avoid critical reasoning, then yes, it is a competitive weakening artifact. Research on cognitive offloading suggests that critical thinking skills can be negatively affected. But if AI is used as a scaffold for deeper thinking—augmenting analytical synthesis, facilitating creative brainstorming, challenging assumptions—it becomes a competitive enabling artifact.
The key, then, is not rejecting AI but designing it and using it in a way that maximizes human engagement rather than replacing it.
Rethinking Education: The End of Memorization?
Much of traditional education is based on the assumption that cognitive artifacts should be minimized. Students are often penalized for using calculators, writing aids, or external references. But if we recognize that cognitive artifacts have often been the engine of intellectual progress, then education should shift its focus:
· Teach cognitive augmentation, in addition to raw skill mastery. Knowing how to leverage AI, data systems, and automation for deeper reasoning is now more critical than memorizing facts.
· Move from knowledge retrieval to knowledge synthesis. Instead of rote memorization, education should focus on problem formulation, interdisciplinary thinking, and metacognition—thinking about thinking.
Civilization and the Future of Intelligence
If we extrapolate this argument forward, we arrive at a provocative conclusion: the evolution of intelligence is not a story of cognitive degradation but of cognitive transformation.
The question of whether AI will make us less intelligent is, in some ways, the wrong one. The better question is whether we will build systems that push us toward greater complexity, abstraction, and synthesis—or whether we will succumb to intellectual complacency.
The fate of intelligence is not about the presence or absence of competitive cognitive artifacts; it is about whether we design them to lift us up or let us atrophy.
David Krakauer’s distinction between competitive and complementary cognitive artifacts captures a real and present danger—there is no doubt that some technologies can erode our cognitive faculties. But history teaches us that cognitive outsourcing is not inherently a loss; it is a trade-off. The true test is whether we ensure that civilization’s cognitive systems don’t just replace thinking but elevate it.
If we get it right, AI and automation won’t be the end of human intelligence. They will be its next great leap.
A special thanks to James Aitken for his editorial suggestions.
References
1. Bohbot, V. D., Bernier, M., & Day, L. B. (2010). Navigation strategies as a predictor of conversion from mild cognitive impairment to Alzheimer disease. Alzheimer Disease & Associated Disorders, 24(1), 27–33
2. Gerlich, M. (2025). AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking. Societies, 15(1), 6. https://doi.org/10.3390/soc15010006
3. Ellington, A. J. (2003). A meta-analysis of the effects of calculators on students’ achievement and attitude levels in precollege mathematics classes. Journal for Research in Mathematics Education, 34(5), 433–463. https://www.jstor.org/stable/30034795
4. Hembree, R., & Dessart, D. J. (1986). Effects of handheld calculators in precollege mathematics education: A meta-analysis. Journal for Research in Mathematics Education, 17(2), 83–99. https://www.jstor.org/stable/749255
5. Javadi, A.-H., Emo, B., Howard, L. R., Zisch, F. E., Yu, Y., Knight, R., Silva, M. A., & Spiers, H. J. (2017). Hippocampal and prefrontal processing of network topology to simulate the future. Nature Communications, 8, Article 14652. https://doi.org/10.1038/ncomms14652
7. Luckin, R., Holmes, W., Griffiths, M., & Forcier, L. B. (2016). Intelligence unleashed: An argument for AI in education. Pearson Education. https://edu.google.com/pdfs/Intelligence-Unleashed-Publication.pdf
8. Maguire, E. A., Gadian, D. G., Johnsrude, I. S., Good, C. D., Ashburner, J., Frackowiak, R. S., & Frith, C. D. (2000). Navigation-related structural change in the hippocampi of taxi drivers. Proceedings of the National Academy of Sciences, 97(8), 4398–4403. https://www.pnas.org/doi/10.1073/pnas.070039597
9. Risko, E. F., & Gilbert, S. J. (2016). Cognitive Offloading: How the Internet Is Changing the Way We Think. Trends in Cognitive Sciences, 20(9), 676-688.
10. Spiers, H., & Jung, M. (2016). Sniffing out routes in virtual reality: The impact of navigation instructions on the brain. Presentation at UCL Institute of Behavioural Neuroscience.
11. Turner, C. (2022). Neuromedia, cognitive offloading, and intellectual perseverance. Synthese, 200(2), 66
12. Weisberg, S. M., et al. (2014). Gender differences in mental rotation and spatial navigation: A review. Behavioral and Brain Sciences, 37(3), 392–393. (Provides overviews on navigation strategies and brain regions involved.)