Generative artificial intelligence and the marginalization of minoritized knowledges in higher education: the case of disability

URL SCAN: Generative artificial intelligence and the marginalization of minoritized knowledges in higher education: the case of disability

FIRST LINE: Submitted on 26 May 2026.

TEXT ANALYSIS

1. The Dissection

This paper performs a social-justice framing of AI's epistemic harms within higher education, using disability as the primary case study. It argues that:
- LLM training data is Anglophone/Western-centric (epistemic coloniality)
- Disabled persons face double marginalization: excluded from design AND confined to reductive stereotypes
- A "hybridization" of researcher and machine might preserve epistemic plurality
- Algorithmic correction is acknowledged as "palliative" only

The paper draws on educational sciences, critical technology studies, and disability studies — three fields that share a methodological inability to model structural economic displacement. They can identify the wound. They cannot read the vital signs.

2. The Core Fallacy

The fundamental error is treating epistemic marginalization as the primary mechanism of harm, when the Discontinuity Thesis identifies a more terminal process: productive economic obsolescence.

The paper treats disability's problem with AI as one of representation, inclusion, and data fairness. It assumes that better datasets, more inclusive design processes, or hybrid researcher-machine workflows can preserve meaningful epistemic plurality. This is category error at scale.

The actual threat model is not that disabled voices are erased from knowledge production. The threat model is that knowledge production itself becomes economically irrelevant to the majority of humans — disabled or not — because the labor market feeding into wage/consumption circuits collapses. Even a perfectly inclusive AI that centers every minoritized epistemology still automates the cognitive work that currently employs the people generating those epistemologies.

The paper is diagnosing the infection while the patient has already entered systemic organ failure.

3. Hidden Assumptions

• Assumption 1: Higher education remains a viable economic institution. The paper assumes universities will continue as meaningful sites of epistemic production. Under DT logic, the institution that produces and validates knowledge faces direct automation displacement — the credentialing and knowledge-production functions are both in the blast radius simultaneously.

• Assumption 2: Inclusive design interventions can outpace deployment velocity. The paper gestures at "hybridization" as a potential solution without modeling the competitive dynamics. Inclusive redesign of AI systems requires resources, coordination, and political will — all of which move slower than capital flight toward automation efficiency gains.

• Assumption 3: Epistemic plurality has economic value under the incoming regime. The paper implicitly assumes that diverse knowledges will be preserved because they have intrinsic worth. Under DT logic, value flows toward what automates productive output. Epistemic diversity becomes a luxury affordance of the Sovereign class, not a right distributed through institutions.

• Assumption 4: The "double marginalization" framing centers the right problem. Marginalization implies exclusion from participation. The DT framing suggests the more accurate description is functional irrelevance — not that disabled persons are excluded from AI systems, but that the economic systems AI enables may not require human cognitive labor at all.

4. Social Function

Classification: Progressive Academic Prestige Signaling + Structural Blunting

This paper does what the critical technology studies / educational sciences literature does consistently — it maps the injury in exquisite, theoretically sophisticated detail while systematically avoiding the prognosis. The work is honest within its frame, but the frame is calibrated to a world where institutional reform is still a viable response vector.

It performs several functions simultaneously:
- Generates academic capital for the author(s) within a discourse community
- Provides intellectual cover for reformist interventions that cannot scale
- Redirects concern about AI away from economic structural death toward cultural-representational harm
- Allows the reader to feel critically engaged without confronting the possibility that inclusion, diversity, and epistemic justice are all lag-level interventions against a P1-P3 mechanical collapse

The paper is not wrong about the marginalization it documents. The data is probably accurate. The failure is one of scope: it has diagnosed a serious wound while the body is in cardiac arrest.

5. The Verdict

The paper identifies real phenomena with genuine precision but fails at the level of systemic threat assessment. It treats epistemic exclusion as the primary pathology when the actual pathology is productive economic displacement of human cognitive labor itself.

Under DT mechanics:
- The "hybridization" proposed is a lag defense at best — a palliative intervention in a terminal structural transition
- Disability studies and critical tech studies as fields are high-risk for irrelevance under AI-reordered institutions, because the institutions they critique are themselves facing structural dissolution
- The framing of "epistemic coloniality" is accurate but insufficient — what training data reinforces is secondary to the fact that the knowledge production economy itself is being automated into obsolescence

Survival utility of this paper's framework: Near-zero. It names the wound accurately. It offers no viable path through it. It is a document produced by intelligent people describing a burning building while the fire suppression system is already offline and the building's load-bearing walls are structural foam.

The irony is precise: papers like this are themselves among the first casualties — academic outputs from disciplines whose institutional funding bases will evaporate as universities face enrollment collapse driven precisely by the AI transformation this paper critiques.

SCORE: The paper is a well-constructed autopsy of a symptom. It is not, and does not pretend to be, a diagnosis of the disease.