Dissertation Proposal Companion · Chapters One to Three
When a machine can produce the visible artifact of learning without the interpretive and authorial work that artifact once signaled, what should schools deliberately keep hard? This is a visual walk through the argument, the framework, and the study design.
ScrollGenerative AI has become routine in K-12 work: planning, writing, tutoring, feedback, and assessment. The familiar questions ask whether students may use it and how to catch cheating. The deeper question is what happens to learning when a polished artifact can be produced without the cognitive, rhetorical, and authorial labor education is designed to develop.
Student work can become more polished while student understanding becomes less clear.
This is not reducible to academic dishonesty. It is a learning-science problem about the work that should connect task, process, artifact, and understanding. And it is an equity problem, because the same tool that bypasses learning for one student can remove a real barrier for another.
Adoption is outpacing institutional coherence. That gap is the ground this study works.
Media ecology treats each communication environment as a milieu that restructures consciousness, not a neutral channel. Ong charted the transformation from orality to literacy to broadcast. The dissertation adds two stages the broadcast model could not anticipate. Select a stage.
If these assumptions hold, today's media are only an intensification of secondary orality. If they break, new categories are required. Generative AI breaks all three.
Was: across every stage, symbols originated in human consciousness.
Now: neural systems produce essays, analyses, and images with no human authorship in any traditional sense. Work can read as understanding without any understanding in its making.
Was: two viewers of one channel saw the same content.
Now: engagement-optimized curation is opaque and personalized. Students arrive having inhabited different informational worlds tuned to prior behavior.
Was: the book on the shelf, the broadcast from the station.
Now: a continuously tailored environment reflects a mind's own patterns back to it, replacing productive disequilibrium with a self-reinforcing loop.
Tertiary algorithmicity names a media environment in which algorithmic systems both curate and generate symbolic content, rendering human authorship optional at scale. This is a qualitative rupture, not a faster version of what came before. Three pressures follow, each endangering a human capacity education depends on.
Stalder's account of algorithmicity describes cultural processes pre-arranged through algorithmic filtering and curation. Tertiary algorithmicity retains that curatorial dimension and extends it to symbolic generation. Algorithmic systems do not only arrange human-created content; they originate symbolic content at scale.
Cognitive labor moves from storage to generation. Synthesis, argument, and meaning-making that once had to happen in the learner are offloaded to an external system.
Communicative environments flood with fluent synthetic discourse. The origin of content becomes uncertain, and simulated dialogue rarely resists, disagrees, or holds anyone to account.
Claims are severed from a claimant. Text can be produced with no consciousness standing behind it and no person bearing the intellectual risk of having committed to it.
A near-term escalation, agentic AI, is treated as an intensification within this same condition rather than a fourth stage. It moves from optional authorship toward optional action. The participant-facing inquiry remains centered on generative AI as educators meet it now, while the nonparticipant comparison retains at least one bounded, prespecified agentic-artifact task.
Learning science supplies the reason. Desirable difficulties, retrieval practice, generation, cognitive load management, and productive failure all show that certain kinds of struggle are conditions of learning rather than obstacles to it.
The felt ease a tool provides is precisely the fluency that research identifies as a poor proxy for learning. Generative AI raises the specific risk of unproductive success, a correct-looking performance produced without the understanding the performance normally indicates. If the environment defaults toward removing struggle, and struggle is constitutive of durable understanding, then the educational response must deliberately preserve the difficulty learning requires.
Chapter Two builds a sequence of warrants. Each tradition performs a different job, and none alone establishes the framework. The defense task is to explain the chain without making the framework look inevitable or using the literature only to confirm an argument already chosen.
Ong, McLuhan, Postman
Explains why communication technologies reorganize consciousness, memory, authorship, and public knowledge.
Stiegler, Kittler, Hayles, Zuboff
Explains how technical systems shape attention, distribution, cognition, institutional power, and the conditions of encounter.
Bjork, Kapur, Sweller
Supplies the warrant for why effort, generation, retrieval, and productive failure matter for durable learning.
Literacy, rhetoric, embodiment
Shows why a fluent artifact can no longer be treated as sufficient evidence of understanding, ownership, or accountability.
Adoption, policy, equity, readiness
Establishes that schools already face the problem and identifies the empirical gap the proposed study will investigate.
Stiegler gives Chapter Two a vocabulary for technical memory and attention that Ong does not develop in the same way. He remains a complementary bridge rather than the developmental anchor.
“Ong helps me explain the historical restructuring of consciousness across media environments. Stiegler helps me explain what changes when memory, attention, and symbolic traces are technically organized. I use Stiegler to deepen the meaning of the transition, not to replace Ong's stages.”
Books, photographs, recordings, and other technical objects preserve traces that consciousness can reactivate. Generative AI complicates this relation by producing retention-like text, image, audio, and video that need not originate in lived experience.
Bridge: existential abstraction, authorship, evidence, and warranted meaning.
A technical system can expand access, memory, expression, and participation while also weakening attention or displacing capacities. This supports a technoskeptical position rather than a ban-or-embrace binary.
Bridge: productive versus exclusionary friction.
Algorithmically tailored environments reflect prior behavior back to the learner. The concern is not only distraction, but whether the environment reduces encounters with resistance, difference, and the disequilibrium through which thinking can change.
Bridge: noetic displacement and infrastructural friction.
Media archaeology keeps hardware, data systems, platforms, and institutions visible. It strengthens the infrastructural dimension, but does not provide Ong's developmental account of consciousness.
Distributed cognition challenges any simple claim that offloading is inherently harmful. Chapter Two must therefore distinguish productive human-tool assemblages from delegation that bypasses the learning goal.
Generative AI remediates the essay, search engine, tutor, and dialogue. This explains relations among media forms, while the dissertation's central question remains what those forms do to consciousness and learning.
Media ecology diagnoses the altered environment. Learning science explains why removing generation, retrieval, revision, and productive failure may matter educationally. One tradition cannot substitute for the other.
Defense answer: Ong remains the anchor because the study needs a developmental account of how dominant communication environments reorganize noetic life. Stiegler deepens technical memory and attention; Hayles complicates the argument through distributed cognition; Kittler foregrounds material and institutional infrastructure. Their differences discipline the framework rather than disappearing into it.
Defense answer: No. The proposal treats technologies as environments with affordances, incentives, and pressures, not as autonomous causes with inevitable effects. The study investigates how educators and institutions interpret and redesign those conditions. Pedagogical and infrastructural friction are precisely claims about human and institutional agency.
Defense answer: Chapter Two preserves augmentation, accessibility, distributed-cognition, and anti-determinist counterarguments. Chapter Three then treats the framework as interpretive vocabulary whose fit, limits, and failures must be examined empirically. The framework is the object of disciplined scrutiny, not the conclusion the study must reach.
Friction here is not energy to minimize. It is the resistance durable learning requires. Three learner-facing dimensions name the human capacities the three pressures endanger. Beneath them, infrastructural friction is the institutional condition that enables or constrains those dimensions. Naming this layer as friction is deliberate. The institutional environment is not a neutral backdrop; it actively constrains or enables whether learner-facing friction can be sustained. Select a dimension.
An argument for preserving difficulty becomes an argument for preserving inequity unless it asks whose learning a given difficulty supports and whose participation it blocks. The framework carries a cross-cutting distinction.
Difficulty that builds capacity: the generative cognitive work, real dialogue, authorship, and accountability through which understanding forms.
Keep it. A student drafts claims by hand before any AI use, then defends the thesis in conference.
Difficulty that blocks access or participation without contributing to learning, often treating one cognitive profile as the unmarked norm.
Remove it. A multilingual learner uses translation to understand directions before starting the real task.
The same tool use can reduce both frictions at once. The question is never whether AI was used, but what work the tool removed and whether that work was relevant to the learning goal.
Because that judgment is situated and no policy can fully specify it, the study investigates how educators actually draw the line in practice rather than assuming the framework resolves it in advance.
This is the study's scenario card sort, from Appendix F. Place each case on two dimensions. There are no correct answers; the point is the reasoning. Set both dimensions on a card to reveal the question the framework asks about it.
The proposal does not test the framework as a fixed set of propositions. It asks how participants make sense of changes in effort, dialogue, authorship, judgment, participation, policy, access, and governance. The design is qualitative-dominant convergent mixed methods and is grounded in constructivist qualitative inquiry.
Select a question to trace its current participant, evidence, analysis, and integration alignment from the July 24, 2026 Chapters 1–3 proposal.
Evidence is analyzed according to its purpose and then brought together to clarify convergence, divergence, expansion, or silence. AI-generated texts and agentic artifacts remain analytically distinct and are never treated as evidence about human experience. The comparison remains open to outputs that display situated judgment, calibrated uncertainty, institutional memory, or productive counter-friction, so it can challenge as well as support the framework. AI may assist with clerical work, but the researcher remains responsible for interpretation.
The proposal defense was passed on August 13, 2026. Interview-question wording is being revised for committee review, followed by a planned three-educator rehearsal to assess conversational flow and an update to the IRB materials. IRB approval, formal recruitment, main-study data collection, coding, analysis, and findings remain future gates.
The methods argument can be explained in three connected moves. Constructivism explains how meaning is understood, sensemaking explains what the questions ask, and thematic analysis explains how patterns are developed across accounts.
Meaning is situated and co-constructed through language, experience, role, and context.
Participant accounts are interpretations, not transparent reports of a single objective reality.
Participants notice cues in ambiguous conditions, interpret them through available frames, form plausible accounts, and act on those accounts.
The questions invite meaning rather than require predetermined conclusions.
First-cycle coding stays close to participant language. Second-cycle analysis develops themes across roles and sources.
The framework sensitizes attention but remains open to revision.
“I use constructivist qualitative inquiry because the study asks how people make meaning under ambiguous educational conditions, while the qualitative-dominant mixed methods design keeps interviews and other open-ended evidence central and uses quantitative evidence for supporting context.”
Code participant language, actions, and meanings closely while writing reflexive memos.
Develop themes across roles and sources without forcing all evidence into the framework.
Use peer debriefing, discrepant accounts, transparent memos, and comparison across evidence sources.
Allow inductive themes, divergence, and silence to revise the framework without a mechanical threshold.
Defense answer: The research questions ask how participants make meaning. Constructivist qualitative inquiry, thematic analysis, and a clear evidence hierarchy provide the necessary methodological warrant without adding a methodological label the study does not need.
Defense answer: The methods have different jobs. Open-ended evidence addresses participant meaning directly. Closed-ended survey items and national datasets provide supporting descriptive and structural context rather than validating participant interpretations.
Defense answer: They are a participant group included in the purpose, RQ1, sampling, and qualitative analysis. Their retrospective accounts provide a learner perspective on teaching and learning under generative AI.
Defense answer: They are a distinct nonparticipant comparison source within the study. The comparison retains at least one bounded, prespecified agentic-artifact task, is analyzed separately, and is never treated as evidence about human experience or sensemaking. If scope must be reduced, systems, task count, or repeated runs are scaled before this component is removed.
Defense answer: The evidence hierarchy and scaling rule keep the study feasible. Role-based practitioner interviews and the educator survey form the analytic core. Retrospective university-student accounts, documents, national datasets, and the nonparticipant artifact comparison supplement that core and can be scaled first. At least one bounded agentic-artifact task remains in the comparison.
This companion is one surface in a larger set of public sites, instruments, data, and games that support the dissertation. Follow any thread.
The public entry point to the whole proposal and its companions.
The argument's lineage: stages, traditions, framework, and a searchable reference library.
The media-ecology stages and the three pressures, in depth.
The friction framework and its dimensions for a practitioner audience.
A workspace for rehearsing claims, methodology, and committee questions.
The peer-reviewed K-12 companion in i.e.: inquiry in education (Miner, 2026b).
Reproducible analysis of the RAND/Gallup and NCES structural-context data (Miner, 2026a).
Walk Chapter Two as seven stations, with badges and a source codex.
Live the memory work of a primary oral culture.
Redesign a task to keep the struggle that matters.
Friction-preserving inquiry for the history classroom.
Writing, consulting, and the broader body of work.