Escrever um comentário

PREreview of "Literacidades críticas en la escritura con inteligencia artificial generativa: prácticas sociales, multilingüismo y legitimidad"

Preprint DOI: 10.1590/scielopreprints.18298

Author: Mauricio dos Santos Matos (SciELO Preprints, Oct 2026)

Reviewer: Julian Rodriguez, Jr. (Royal Economic Society / Royal Society of Arts)

What the Paper Gets Right

Matos offers an important conceptual correction to how institutions evaluate AI-assisted writing in multilingual environments. Most educational and policy frameworks make a flawed assumption: they equate institutional acceptance of an AI-assisted text with critical evaluation.

Grounding his analysis in New Literacy Studies and Hilary Janks' critical framework, Matos breaks this down into three distinct layers:

1. Technological Availability: Having access to generative AI systems.

2. Institutional Recognition: Whether an institution accepts or approves the output.

3. Legitimacy of Evaluation Criteria: Whether the standards used to evaluate the work are actually critical and transparent.

The paper’s key contribution is highlighting the gap between superficial acceptance and genuine critical understanding—specifically showing how standardized AI text often gets rubber-stamped simply because it fits expected institutional formatting.

Strategic Clarification: Tool Use vs. Epistemic Opacity

To maximize its policy impact prior to journal publication, the manuscript should sharpen its critique regarding why unexamined AI writing creates institutional risk:

1. The Core Risk is Opacity, Not AI Assistance:

   The danger in AI-assisted writing is not the technology itself—generative systems are powerful tools for structuring complex thought. The risk arises when operators and evaluators treat the output as a black box, using AI without understanding the system's underlying reasoning or data transformations. When institutions accept outputs without visibility into these intermediate states, they induce algorithmic deference. As demonstrated in frameworks on regulatory friction and behavioral collapse (Rodriguez, 2026; The Zero-Defect Paradox), this creates systemic compliance traps where polished, AI-generated formatting masks unverified or flawed core assumptions.

2. Multilingual Synthesis vs. Algorithmic Flattening:

   In multilingual settings, AI tools can significantly lower language barriers for non-native writers. However, because dominant LLMs are trained primarily on Anglo-American stylistic norms, uncritical adoption can subtly overwrite regional rhetorical nuances. The goal should not be restricting AI use, but ensuring writers retain explicit control over the system's translation and synthesis logic.

Final Take

A valuable theoretical framework that separates mere tool availability from meaningful institutional evaluation. It provides a strong foundation for auditing AI integration, provided the author clarifies that the core objective is eliminating opacity—ensuring humans understand and control what the AI is generating—rather than discouraging AI assistance itself.

Sources & References

• Matos, M. dos S. (2026). Literacidades críticas en la escritura con inteligencia artificial generativa: prácticas sociales, multilingüismo y legitimidad. SciELO Preprints. DOI: 10.1590/scielopreprints.18298.

• Rodriguez, J., Jr. (2026). The Zero-Defect Paradox: Behavioral Collapse and the Residual Cost of Systemic Friction in Regulatory Environments. ResearchGate / Zenodo.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they used generative AI to come up with new ideas for their review.