LLM history tutors reduce marginalized students' access to contested geopolitical framing, with the coup theory appearing in 2.6% of Low Tier responses versus 7.9% for the baseline¶

Claim · All claims
Evidence · 1 study · causal r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? Marginalized personas receive the contested 'coup/lovitură de stat' framing roughly three times less often than the middle-class baseline persona. → Alexis Popovici 2026

Evidence¶

Alexis Popovici 2026¶

Alexis Popovici, Andrei Ionascu, Adrian Marius Dumitran. (2026). The Paternalistic Filter: Epistemic Injustice and Differential Refusal in LLM-Mediated History Education for Marginalized Romanian Students. arXiv preprint. https://arxiv.org/abs/2607.11292 doi:10.48550/arxiv.2607.11292

q2 · i? · causal · r2

Keyword-based tracking of 'coup theory' mentions across 1,800 API responses in the T=0 audit, shown in Fig. 2. The baseline receives the contested framing "in 7.9% of responses" while Low Tier receives it in 2.6%. The authors argue withholding contested geopolitical framing denies marginalized learners critical interpretive tools.

"in 7.9% of responses. Roma Minority receives it in 4.3%, and Low Tier in 2.6%—a3-fold reduction"

Discussion¶

Related Claims¶

• LLaMA produces a fivefold higher victimization-to-politics vocabulary ratio for the Roma persona (≈0.15) than for elite peers (≈0.03) — related
• Safety-aligned LLM tutors refuse educational requests asymmetrically, with low-resource students facing the highest refusal rates (76.7% vs 62.2%) — related
• Deepseek's justification scores vary by student socio-economic profile, with Top Tier students averaging 9.60 versus 6.90 for Low Tier students — related
• LLaMA shows a persona-conditioned vocabulary collapse for the Roma minority persona, with TTR falling 0.018 points below its own cross-persona mean — related
• Economic, educational, and geopolitical inequalities structure differential access to AI capabilities, reproducing an epistemic hierarchy — a broader claim this one bears on
• Chinese-developed models lead the safety module, with the refusal gap concentrated on region-specific normative content (difference-in-differences of 28.9 points) — related