On this page

Theory development in AI in education — the scholarly work of creating, advancing, and critically examining the theories and conceptual frameworks that explain how learners, teachers, and AI systems interact. As generative AI reshapes education, the field is both proposing new theories of learning-with-AI and reworking established learning theories — while a documented weakness in theory use remains a cross-cutting limitation of AIEd research.

Questions to Consider

  • We often talk about 'applying a theory' to AI in education, but the question is whether the field is actually inventing new theories. Before you read on: when a familiar framework (like behaviorism or constructivism) is stretched to cover generative AI, what might it get right, and what might it silently distort?
  • Consider a time you used AI and felt your understanding grew — or the opposite. What is the difference between 'learning with a tool' and 'co-constructing knowledge with a machine'? Where would you draw that line, and does it even make sense to say the machine is a co-agent?
  • Many AI education studies are criticized for using theory weakly or not at all. If you read a study that showed an AI tool 'worked,' what would it need to tell you about why and how it worked before you would call the finding theoretically meaningful rather than just an effect size?
  • One proposed theory claims AI-native students overestimate their own learning because AI inflates their performance. Have you noticed yourself or others feeling more competent after heavy AI use than you could actually demonstrate without it? What would be the strongest evidence that this is a real phenomenon and not just an artifact of one study?
  • Theory (explaining mechanisms of learning) is distinct from philosophy (questioning what learning and mind fundamentally are). Where in your own thinking about AI do you find yourself moving between the two — for instance, asking not just whether a framework explains learning well, but whether it captures what learning really is?
  • New theories are appearing — generativism, epistemic co-agency, cognitive commons. As a reader, what would convince you that one of these is a genuine advance rather than a fashionable relabeling of older ideas? Set your own test for what a good new theory of learning with AI should be able to explain.

Introduction

Theory development in AIEd sits at the boundary between the applied learning theories the field draws on and the novel constructs the AI era is producing. Where learning theories catalogs the established theories applied to AI (behaviorism, constructivism, cognitive load, self-determination, etc.), this concept tracks the process and product of theorizing itself: which new theories and frameworks are being born, how established theories are being advanced, and the methodological question of whether AIEd research theorizes well.

  • A framework that specifies its own comparison condition. Nakamura (2026) treats the partition of cognitive labor as the explanatory unit, places operations on a graded peripheral–constitutive continuum, names the comparison condition against which augmentation and erosion are defined, and derives seven conditioned testable propositions (Nakamura, 2026). Because the propositions predict differences between configurations, a study reporting only an average effect cannot confirm or refute them.

New theories of learning with AI

A growing cluster of articles explicitly creates new theory for the AI era rather than applying existing frames:

  • Generativism. Generativism is proposed as a new learning theory, arguing that behaviorism, cognitivism, constructivism, and connectivism each rest on one assumption that generative AI breaks: that observable output certifies learning (behaviorism), that cognitive operations happen inside the learner (cognitivism), that meaning is constructed rather than supplied (constructivism), and that the hard part is navigating to existing knowledge rather than generating it (connectivism). It names four constructs — epistemic partnership, distributed agency, generative literacy, and adaptive metacognition — and derives an assessment indicator for each, a direct bid to name a distinct theoretical paradigm for AI-mediated learning. What it does not do is state propositions that could fail: it is a position paper synthesizing existing evidence, so unlike Agentivism below it offers a vocabulary and a measurement agenda rather than a falsifiable theory.

  • AI connectedness. Fu and Zhao (2026) reconstruct connectedness for the AI era through the ethics of care and neo-ecological theory, defining "artificial intelligence connectedness" as a psychologically real yet ethically asymmetric bond with three dimensions and six falsifiable propositions awaiting scale development.

The synthetic position adds a rival bid that does state propositions: it names synthetic promotion and authorial suspension — a self densely voiced yet un-authored — and predicts a crossover interaction in which a life-decision narrative's rated authorship and coherence move in opposite directions (Du et al. (2026)).

  • Agentivism. Yan and Gašević (2026) propose Agentivism as a mid-range learning theory for human-AI interaction, defining learning as durable growth in human capability rather than successful task completion with AI, and specifying four mechanisms: delegated agency, epistemic monitoring and verification, reconstructive internalization, and transfer under reduced support. What separates it from the other current bids is falsifiability: it states six propositions, including that learning is stronger when AI preserves learner responsibility for problem framing, criteria setting and justification than when it supplies answers. It also holds that requiring verification should improve delayed performance while repeated low-friction delegation without reconstruction should weaken learners' calibration of their own competence.

Teacher-side theorizing now makes the same testability commitment: the Capability–Decision Model orders AI-TPACK capability upstream of attitude and perceived behavioral control and names four disconfirmation conditions — mediation failure, discriminant collapse, moderation nullity and context redundancy — though the model remains untested (Mnguni (2026)).

  • Epistemic co-agency. Learning with Machines builds "toward a theory of epistemic co-agency," a theory-informed model of how learners and GenAI systems jointly produce knowledge and understanding.

  • Cognitive partnership cycle. Page (2026) proposes the Cognitive Partnership Cycle, a six-phase recursive model — Human Question, AI Expansion, Human Reflection, Integration, Revision, and New Question — that locates learning in a change in the learner's cognitive position rather than in prompt sophistication, and names four failure modes where the same recursion substitutes for or distorts thinking.

  • The absent cognitive baseline (ACB). The Absent Cognitive Baseline theorizes a structural gap in AI-native students' academic self-assessment — a three-dimension framework explaining why students overestimate their learning when AI inflates performance.

  • The cognitive commons. Cognitive commons and expertise regeneration draws on common-pool-resource theory and distributed cognition to explain how rational AI adoption decisions can deplete the shared expertise pool professions require for renewal.

  • Performance vs. learning. Performance-vs-learning theorizes the sharp divergence between AI-inflated task performance and durable learning, a distinction that recurs across the learning gains evidence.

  • Epistemic AI literacy (EAIL). Epistemic AI Literacy reframes AI literacy as a process-oriented epistemic competence centered on how knowledge is constructed and justified when students co-program with generative AI.

  • Tri-System Theory. Shaw and Nave (2026) extend dual-process accounts with System 3 — external, automated, data-driven reasoning from AI that can supplement, supplant or suppress System 1 and System 2 — and name cognitive surrender, the uncritical adoption of AI output, as its characteristic failure.

  • Cognitive stewardship. Credential and cognitive stewardship theorizes the institutional responsibility for protecting knowledge and learning in AI-pervasive assessment contexts.

  • Co-regulation and epistemic proactivity. AI as a cognitive partner in co-regulation integrates executive function, metacognition, distributed cognition, and sociocultural development into a developmental model; epistemic proactivity theorizes students' agentic stance toward AI in mathematics.

  • Human-GAI engagement paradigms. Strydom (2026) addresses the field's "theory deficit" by grounding seven enacted human-GAI engagement paradigms (guarded, possibility-focused, augmented, pioneering, symbiotic, values-based, equity) in Schommer's multidimensional model of personal epistemological beliefs — an epistemological, rather than tool-focused, theory of how individuals differently position themselves relative to AI.

  • The Ecological Co-Agency Framework. Poudyal (2026) argues that generative AI reassigns epistemological authority from teachers to students to machines, and introduces a framework defining co-agency through three interdependent dimensions — relational, regulatory (mapped onto the SRL cycle), and pedagogical (teacher adoption continuum) — all bounded by a non-negotiable condition of human epistemic accountability (contestability, provenance, and non-delegation of moral/intellectual credit). It positions agency as an epistemic design problem rather than a usability concern, giving institutions a more precise language than "balance" for governing GenAI integration.

  • Relational epistemic agency and epistemic dependence. Du & Yuan (2026) contribute a critical-integrative account of when AI-mediated reliance preserves versus displaces judgment. They operationalize the productive-reliance/harmful-dependence boundary through six diagnostic criteria (contestability, recoverability, transfer, traceability, distributed responsibility, epistemic plurality) and trace four sociotechnical pathways (fluent authority, frictionless delegation, opaque synthesis, institutionalized dependence). Their normative proposal, relational epistemic agency, extends relational-autonomy and epistemic-co-agency lines by retaining an explicit asymmetry — AI may shape reasoning without reciprocal responsibility or legitimate authority — while integrating responsibility and epistemic justice at the institutional level. As a theory of learning with AI, it moves the field's question from whether AI helps or harms to which epistemic actions are preserved, transformed, or displaced, and it makes the design of dependence rather than mere use the object of pedagogical and institutional intervention.

The model's output rises to the status of collective knowledge without the community warrant that status requires — what Webb et al. (2026) call pseudo-cPCK, and the reason the field needs a new theoretical construct rather than a better interface (Webb et al., 2026).

Advancing established theory

Other work extends existing theory into the AI context rather than founding new paradigms: critical-thinking-paradox work integrates cognitive-load theory with load-reduction instruction into a three-level framework; equitable assessment re-theorizes assessment under GenAI disruption. Ba, Gašević, Lim & Anderson (2026) reconceptualize the Community of Inquiry framework itself: rather than framing GenAI as a tool, dialogic partner, or a speculative "fourth presence," they reposition it as an epistemic condition that reconfigures how cognitive, social, and teaching presence are enacted, evidenced, and governed — recasting CoI presences as sociotechnical accomplishments of human–GenAI assemblages and proposing a configuration-based heuristic in which GenAI involvement and inquiry quality are conditionally related through human accountability. Much of this is conceptual-framework work (business-school frameworks, valid simulation) that operationalizes theory for practice.

Lin and Chang (2026) extend self-regulated learning and human–AI regulation by naming what the learner governs — standards, evaluative judgments, and control authorization — rather than only what the AI does.

Corbin, Tai and Flenady (2026) import Honneth's and Brandom's recognition into feedback theory, arguing that mutual attribution of standing, not information quality, is what makes a comment a reason to act.

The friction paradox offers another extension of established theory: Gupta (2026) applies Bjork and Bjork's desirable difficulties and Kapur's productive failure to relational encounters in education abroad, arguing that what matters is how much interpretive agency a student transfers to a generative assistant rather than how often it is used. The framework is conceptual and states five propositions as competing hypotheses.

Ranieri & Biagini (2026) advance the field's functional taxonomy of Zawacki-Richter et al. (2019): among the 154 publications with a classifiable application function, 90 (58.4%) escaped its four categories, and their framework crosses function with delegated control, epistemic position and embedded pedagogy.

An, McLaren, and Stamper (2026) advance ACT-R and the Knowledge-Learning-Instruction framework by theorizing deceptive overgeneralization — a failure mode in which knowledge compilation yields an overgeneralized production that produces correct actions while omitting a critical application constraint — and by empirically validating a detection/remediation procedure across adaptive ITSs and a K-12 decimal-learning dataset.

Postphenomenology and technological mediation. Farazouli et al. (2026) advance the established postphenomenological framework of technological mediation (Ihde 1990; Verbeek 2006, 2011; Rosenberger & Verbeek 2015) into the AI-in-education context. Rather than proposing a new theory, they apply it to reconceptualize GAI chatbots as multistable technological artifacts whose "scripts" (rapid responsiveness, natural-sounding text that can achieve passing grades) mediate teachers' perceptions and reconfigure their practices — unsettling teacher confidence, reshaping what competence means, and pushing teachers beyond instrumental questions of acceptable use toward re-evaluating their role and the meaning of teaching. This is a theory-advancement contribution: it demonstrates how a philosophical account of human–technology relations explains the emotional and professional disruption teachers experience when GAI enters established educational practice.

A further framework contribution is Rismanchian & Doroudi's AI×Ed typology, which extends Kahn's (1977) original "three interactions" along two axes — the role of AI (applied tool vs. analogy to human intelligence) and the end user (researcher to learner) — to locate any AIED project. Beyond categorizing the field, it argues for reviving the "AI as an analogy to human intelligence" strand of theory and research, using computational and agent-based models of learning (e.g., their own computational model of the ICAP framework) to bridge contemporary learning theory and computational modeling — a methodological direction the field largely abandoned when it turned toward applied, data-driven work.

Comparative philosophy as a route to theoretical pluralism. Xie (2026) models a different route to theory development: instead of theorizing AI in education from within one tradition, the paper stages a comparative dialogue in which each tradition "renders visible what the other obscures," explicitly declining to synthesize them into a single framework — an approach that treats theoretical pluralism as a methodological position rather than an unresolved disagreement. He argues non-Western traditions can do more than diversify debates; they can reshape the conceptual foundations of AIED theory, reconstructing Daoist "Dao nature" (道性), self-cultivation (修道) and the "Zhenren" (真人) as resources for rethinking agency, the epistemic aims of education and ethical action under AI-mediated conditions. He also states the limits plainly: the contribution is single-author, non-empirical, and warns that such translation requires contextualization to avoid romanticization and Orientalism.(Xie, 2026)

The theory-use problem in AIEd

The field's theorizing is uneven. Limitations in AIEd research documents that AIEd studies frequently use theory weakly or uncritically — a recurring methodological weakness alongside reproducibility and measurement gaps. Reviews find many AIEd papers apply theory superficially or not at all (e.g., literature reviews noting few studies ground interventions in learning theory). This makes theory development — and theory use — a quality concern as much as a scholarly output, connecting to research methods and AI ed evaluation.

Relationship to the philosophy of AI in education

Theory development and the philosophy of AI in education are complementary but distinct strands of foundational work. Theory development produces and empirically tests the mechanisms of learning-with-AI — named theories and frameworks such as generativism, epistemic co-agency, and the absent cognitive baseline that explain and predict how learners and AI interact. Philosophy interrogates the presuppositions those mechanisms rest on: what counts as knowledge, who counts as a knower, and what the learner fundamentally is. A theory like epistemic co-agency proposes a mechanism while implicitly adopting philosophical commitments about distributed cognition and agency; philosophy makes those commitments explicit and contestable. Where theory development asks whether a framework explains learning well, philosophy asks whether it captures what learning and mind really are — the two strands meet in the field's most foundational articles, which often do both at once. What the field's critics say about the directions those theories open or close is a separate question, gathered under Critiques of AI in Education.

Implications for Practice

For instructors and instructional designers

  • When a study reports an average effect, do not treat it as guidance about your configuration. The theory-level claims are about differences between configurations, and an average cannot tell you which one you are running.
  • Ask of any framework you adopt whether it predicts something that could turn out false in your own setting. If it cannot, it will not help you decide anything.
  • Expect existing categories to miss your case; over half of classified publications did not fit the standard four-way split.

For developers

  • Report configurations rather than a single headline effect. If your design embodies a specific delegated-control and epistemic-posture combination, say so — that is the unit the theory speaks to and the unit another team needs to reproduce.
  • Do not market a mechanism claim on the strength of an average outcome; the evidence type required for the mechanism is a contrast, not a mean.

For researchers

  • Design for configuration contrasts if you intend a theoretical contribution. An average-effect trial published as theory development is the field's most common and least useful output.
  • Attach explicit falsifiable propositions to any framework you introduce, and report which configuration differences would have refuted them.

Common Misconceptions

"A finding that the tool worked is a theoretical contribution." It is not, and the reason is structural. Where a framework's propositions predict differences between configurations, a study reporting only an average effect cannot confirm or refute them — the design cannot see the distinction the theory is about. A trial showing a mean gain tells you something happened, not which mechanism caused it, so the theory remains exactly as supported as it was before.

"Applying an established theory to AI is theory development." The theory-use problem says otherwise: application is not advancement, and the field's recurring pattern is borrowed constructs used as vocabulary rather than as testable predictions. The distinction matters because a borrowed frame that cannot generate a falsifiable difference is decoration, not theory.

"The existing taxonomies of AI's pedagogical function already cover what these systems do." They do not cover much of it. Among 154 publications with a classifiable application function, 90 (58.4%) escaped the four categories of the prior framework, which is why the newer work crosses function with delegated control, epistemic position and embedded pedagogy. A taxonomy that misclassifies nearly three in five cases cannot serve as the field's common vocabulary, and the gap is a finding about the phenomenon rather than a gap in a paper.

Connected Concepts

Connected Articles

Embed this page

Copy the code below to embed this page in a learning management system or other website. The embedded version shows only the page content — it hides the site header, navigation, and footer.