A Biological Learning Theory: How We Learn in the Era of AI

Abstract Artificial intelligence (AI) is transforming education, yet its impact on learning remains minimally theorized. Existing pedagogical models rarely address AI’s role, leading to misapplications - such as comparing student performance to AI outputs - without understanding how AI influences learning. This paper proposes a biological theory of learning grounded in autopoiesis and adaptation, reframing learning as a process of biological change driven by energy modulation and the regulation of biological perturbation. We revisit foundational assumptions about human nature and analyze three scenarios: (1) AI versus traditional tools, (2) human-human versus human-AI interaction, and (3) the learner’s relationship with AI. This framework moves beyond debates like constructivism versus behaviorism, offering a deeper understanding of learning as embodied transformation. AI is not merely the context for this framework but the phenomenon that renders it necessary: by producing learning-like outputs without biological learning, AI forces a precision in the definition of learning that no prior technology demanded. Ultimately, we challenge the trend of modeling students after AI systems, advocating instead for educational environments that honor complexity, emotion, and human adaptability. We ground these arguments in science education as the domain where AI’s entry into learning environments is most acute and where the biological dimensions of learning — hands-on experimentation, sensory perturbation, and embodied inquiry — are most clearly at stake.