Why Your Child's Education Shouldn't Be Outsourced to AI: Lessons from Evolutionary Biology
Teaching Can't Be Outsourced
Living in the Umwelt
What is it like to be a tick?
Thomas Nagel's 1974 essay What Is It Like to Be a Bat? posed a profound question about bat consciousness that continues to challenge our understanding of subjective experience. But I want to push this inquiry in a more radical direction: What can the seemingly insignificant tick lying in wait on a blade of grass for a patch of hairless, mammalian skin to pass by tell us about consciousness, meaning-making, and the fundamental differences between biological and computational systems?
I argue that the tick—far from being a mere stimulus-response machine—reveals something essential about how meaning emerges in biological systems that cannot be fully replicated in even our most sophisticated language models. This distinction isn't academic; it might fundamentally reshape our ethical frameworks and technological aspirations.
The Revolutionary Implications of Biological Meaning-Making
The Tick's World: More Than Mechanical
The tick has become a biological cliché, reduced to its three-signal detection system: butyric acid, temperature, and skin texture. But this reductionist view obscures what makes the tick's existence revolutionary.
What matters isn't just that the tick detects these signals, but that these signals matter to the tick. The tick doesn't merely process butyric acid as data—it experiences it as significance. For the tick, butyric acid isn't just a chemical compound; it represents the difference between life and death, between reproduction and evolutionary oblivion.
This is the crucial point that many accounts of sensory systems miss: biological perception isn't just about information processing—it's about meaning generation. The tick's sensory apparatus evolved precisely because it created a meaningful relationship between the organism and its environment. The tick doesn't just detect; it cares about what it detects, because its existence depends on it.
I contend this distinction between detection and meaningful experience represents a significant difference between biological systems and computational ones—though this difference may be one of degree rather than an unbridgeable categorical divide.
Plant Intelligence: Rethinking Cognition Without Brains
Plants push this insight even further by demonstrating that meaning-making doesn't require neurons. Consider the acacia tree that, when browsed by giraffes, produces tannins and emits ethylene gas that signals to neighboring trees, which then increase their own tannin production. This isn't just a chemical chain reaction—it's an evolved system of environmental monitoring and response that creates meaningful relationships between trees and their ecosystem.
Plant biologist Anthony Trewavas argues that we should recognize this as a form of plant intelligence—not metaphorically, but literally. Plants make decisions, remember past events, learn, and communicate. They do this without neurons or central processing, demonstrating that intelligence and meaning-making can emerge from networked, distributed systems that bear little resemblance to animal brains (Trewavas, 2003).
This isn't anthropomorphizing; it's recognizing that our brain-centric definitions of intelligence and meaning have been too narrow. Plants have evolved sophisticated ways of creating meaningful relationships with their environments that don't depend on centralized consciousness as we understand it.
Uexküll's Umwelt: A Theory of Biological Meaning
Jakob von Uexküll's concept of Umwelt provides the theoretical framework for understanding these diverse forms of biological meaning-making. The Umwelt isn't just the set of sensory capacities an organism possesses—it's the world of significance that emerges from the relationship between an organism and its environment.
Uexküll argued that all living beings should be understood as subjects, and that the worlds they inhabit are constituted through their specific ways of perceiving (Brentari, 2015). Each species creates its own meaningful world through its evolved sensory and cognitive capacities. The tick's world of butyric acid and the plant's world of chemical exchanges are not impoverished versions of human reality—they are complete, coherent worlds of meaning organized around the biological imperatives of different forms of life.
Contemporary scholars have continued to develop Uexküll's ideas, with some like Tønnessen (2015) extending the concept to explore inter-species relationships and even incorporating it into ecological philosophy. However, it's important to note that Uexküll's original theory had problematic aspects, including its reliance on a false premise of environmental stability (Tønnessen, 2009), which modern interpretations have had to revise in light of our understanding of environmental change and evolution.
Language Models: The Biological Substrate of Meaning
Language models differ from biological systems not merely in complexity or architecture, but in their fundamental relationship to existence itself. This distinction isn't absolute but relates to how meaning emerges in systems with different relationships to their environment.
Some will object that sufficiently complex systems might develop emergent properties that resemble biological meaning-making. This is a reasonable objection, especially as AI systems like large language models demonstrate increasingly sophisticated capacities for adaptation. However, current language models still lack key aspects of biological meaning-making: they aren't designed to maintain their own existence in response to environmental pressures. They don't participate in the evolutionary processes that have shaped biological meaning-making for billions of years.
The concept of autopoiesis, developed by biologists Humberto Maturana and Francisco Varela, helps clarify this distinction. Autopoietic systems are self-producing and self-maintaining—they continuously regenerate the networks of processes that produced them (Luisi, 2003). As Maturana wrote, living systems are networks of molecular exchanges that continuously interact with their environment while maintaining their identity. This self-maintenance creates a boundary that determines what constitutes the system and separates it from its environment—a functional closure that nonetheless allows interaction with the outside world.
When a language model becomes obsolete, it doesn't experience this obsolescence as meaningful because its architecture doesn't include the autopoietic processes that make survival meaningful for biological entities. The AI doesn't fight for its continued existence or adapt to prevent its termination. While reinforcement learning creates feedback loops that can shape AI behavior, these systems lack the fundamental property of maintaining their own physical structure through metabolic processes.
Recent developments have begun to blur these lines, however. Adaptive AI systems are becoming more capable of self-modification, and some researchers are exploring architectures that incorporate elements of self-maintenance and environmental coupling (IBM, 2024). The distinction between biological and computational systems may be less categorical and more a matter of degree than originally supposed in early formulations of autopoiesis.
Why This Distinction Matters for Technology Design
By grounding our technological development in this biological understanding of meaning, we can create AI systems that work with rather than against the diverse forms of meaning-making that have evolved over billions of years.
Current AI development often attempts to simulate biological intelligence by creating systems that mimic human capabilities—generating text, recognizing images, or making decisions. But this approach misses the fundamental distinction between simulation and participation in meaningful relationships. Instead of trying to build AI systems that pretend to have their own stakes in existence, we should design technology that enhances the meaningful relationships that already exist in biological systems.
For example, rather than developing autonomous AI systems that attempt to replace human judgment in environmental management, we could create tools that extend our ability to perceive and respond to the meaningful signals in other species' Umwelten. AI-powered monitoring systems that detect subtle changes in pollinator behavior or forest health can amplify our awareness of ecological relationships without claiming to understand or replace them.
In healthcare, instead of pursuing AI diagnosticians that attempt to replicate human medical expertise, we might focus on systems that enhance the doctor-patient relationship by filtering and organizing information while leaving meaning-laden ethical judgments to humans with genuine stakes in the outcomes. These systems would respect the unique capacity of human clinicians to understand the meaning of illness in a patient's life story.
These approaches don't treat technology as a replacement for biological intelligence but as a complement to the diverse forms of meaning-making that evolution has produced. By designing AI to enhance rather than simulate meaning, we create technology that serves rather than supplants the richness of biological experience.
Education: Preserving Biological Meaning-Making in an AI Era
The distinction between biological meaning-making and computational pattern recognition has profound implications for educational technology. During adolescence, humans are actively constructing their Umwelt—their meaningful world—through cognitive development and social interaction. This biologically grounded process of making meaning differs fundamentally from AI's pattern recognition.
Neuroimaging studies reveal the biological basis for this distinction. Research from Voss and colleagues (2011) demonstrated that active versus passive learning produces significantly different brain activation patterns. When students solve problems through active engagement rather than passive observation, they develop stronger neural connections in regions associated with long-term memory and conceptual integration. This biological foundation—the way intellectual struggle literally shapes neural architecture—must inform our approach to AI in education.
When an adolescent writes an essay about their family history, they engage in a meaning-making activity with existential stakes. The student's engagement extends beyond completing an assignment; it connects to their identity formation and place within meaningful social networks. This process activates networks in the prefrontal cortex and temporal lobes that strengthen with practice but remain dormant when students merely edit AI-generated text (Peters et al., 2021).
The distinction lies not in whether AI handles "basic" versus "advanced" tasks, but in whether students remain the primary meaning-makers in the learning process. When a student asks AI to generate a complete essay which they merely edit, they bypass crucial neurological processes involved in knowledge synthesis and conceptual integration. They have outsourced learning. However, students can leverage AI for sophisticated cognitive assistance while maintaining their role as the central meaning-maker:
Using AI to provide critical feedback on their arguments forces students to evaluate and defend their reasoning, activating reflective thinking processes in the prefrontal cortex.
Having AI simulate different audience responses helps students develop perspective-taking abilities, engaging social cognition networks that are essential for nuanced communication.
Engaging in Socratic dialogues with AI about their ideas prompts students to articulate, defend, and refine their thinking—cognitive processes that strengthen connections between conceptual understanding and verbal expression.
Asking AI to identify unstated assumptions or logical gaps in their reasoning challenges students to engage in metacognition about their own thinking.
The crucial difference is agency: does the student remain the primary architect of meaning, using AI as a cognitive partner that prompts deeper thinking? Or does the student outsource the fundamental meaning-making process, becoming merely an editor of machine-generated content?
AI should not be banned. Outsourcing can and should be banned.
Educational policies should establish clear boundaries that preserve student agency while embracing AI's potential to enhance cognitive development. AI should not be banned. Outsourcing can and should be banned. This isn't about restricting technology but about designing educational practices that leverage AI to enhance—rather than replace—the biological processes through which intellectual challenge shapes cognitive development. Assessment frameworks should likewise evolve to evaluate not just finished products but the meaningful relationships students develop with content—their ability to connect ideas to lived experience, debate perspectives with emotional investment, and incorporate knowledge into their developing sense of self.
By recognizing learning as a biological process of meaning-making rather than mere information processing, educators can design environments that leverage AI while preserving the existential stakes that make education transformative. The goal isn't to shield students from technology but to ensure they remain the primary authors of their cognitive development, using AI to expand rather than replace their biological capacity for creating meaning.
Beyond the Binary
The tick doesn't just detect butyric acid—it experiences it as meaningful. This seemingly simple fact opens a window into the richness of biological meaning-making that extends from the simplest organisms to complex human societies. By acknowledging the profound difference between detecting and caring, between pattern recognition and existential stakes, we can develop more thoughtful approaches to both technology design and education.
Rejecting computational systems totally is unlikely to achieve its desired outcome, in my opinion, but understanding their relationship to biological meaning-making is central to solving the dilemma. Language models and other AI systems can serve as powerful extensions of human cognition without needing to replicate the autopoietic structures of living systems. By respecting the boundary between computation and participation, we can teach our students how to enrich rather than replace the diverse forms of human biological meaning-making that have evolved over billions of years.
The most promising path forward isn't to continue pursuing ever more convincing simulations of meaning nor to rely on containment policies but to design systems that amplify our capacity to participate in, perceive, and struggle to grasp meaningful concepts and relationships that already exist in the biological world. By anchoring our technological development in an understanding of meaning as emerging from evolved relationships between organisms and environments, we can create systems that serve the flourishing of life in all its forms.
References
Brentari, C. (2015). Jakob von Uexküll: The Discovery of the Umwelt between Biosemiotics and Theoretical Biology. Springer. https://content.e-bookshelf.de/media/reading/L-3934410-0e73fdf2dc.pdf
IBM. (2024). The Top Artificial Intelligence Trends. Retrieved from https://www.ibm.com/think/insights/artificial-intelligence-trends.
James, K. H., & Swain, S. N. (2013). Brain activation patterns resulting from learning letter forms through active self-production and passive observation in young children. Frontiers in Psychology, 4, 567. https://pmc.ncbi.nlm.nih.gov/articles/PMC3780305/
Luisi, P. L. (2003). Autopoiesis: a review and a reappraisal. Naturwissenschaften, 90(2), 49-59. https://constructivist.info/18/1/020.luisi
Maturana, H. R., & Varela, F. J. (1980). Autopoiesis and Cognition: The Realization of the Living. D. Reidel Publishing Company. https://www.abebooks.com/9789027710154/Autopoiesis-Cognition-Realization-Living-Boston-9027710155/plp
Nagel, T. (1974). What Is It Like to Be a Bat? The Philosophical Review, 83(4), 435-450. https://www.sas.upenn.edu/~cavitch/pdf-library/Nagel_Bat.pdf
Peters, S., Lim, S. B., Louie, D. R., Yang, C. L. H., & Eng, J. J. (2021). Brain activation differences during active and passive walking in healthy and stroke populations: A systematic review. Neuroscience & Biobehavioral Reviews, 127, 638–654. https://doi.org/10.1016/j.neubiorev.2021.05.037
Tønnessen, M. (2009). Umwelt Transitions: Uexküll and Environmental Change. Biosemiotics, 2, 47-64. https://scholar.google.com/citations?user=RCBb6dMAAAAJ
Tønnessen, M. (2015). Introduction: The Relevance of Uexküll's Umwelt Theory Today. In Jakob von Uexküll and Philosophy (pp. 1-20). Springer. https://content.e-bookshelf.de/media/reading/L-3934410-0e73fdf2dc.pdf
Trewavas, A. (2003). Aspects of Plant Intelligence. Annals of Botany, 92(1), 1-20. https://www.researchgate.net/publication/10766408_Aspects_of_Plant_Intelligence
Voss, J. L., Gonsalves, B. D., Federmeier, K. D., Tranel, D., & Cohen, N. J. (2011). Hippocampal brain-network coordination during volitional exploratory behavior enhances learning. Nature Neuroscience, 14(1), 115-120. https://pmc.ncbi.nlm.nih.gov/articles/PMC3057495/