AI evidence extraction

Inteligencia artificial generativa, paradigmas en crisis y el futuro epistémico de la investigación

Daniel Andrade-Girón, William Joel Marín Rodriguez, Marcelo Gumercindo Zúñiga-Rojas · 2025

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10

Relevance

2/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.36881/ia2025.3

Methodology & findings

Study design

Qualitative, exploratory-analytic, and transdisciplinary approach combining hermeneutic-interpretive analysis and critical technology studies.

Main result

The study found that "la inteligencia artificial generativa es un actor epistémico disruptivo que desencadena una transformación con respecto a las formas de observar, modelar y validar conocimientos" (generative artificial intelligence is a disruptive epistemic actor that triggers a transformation regarding ways of observing, modeling, and validating knowledge). Additionally, the research demonstrates that "el aporte de la inteligencia artificial trasciende la eficiencia operativa, habilitando nuevas fronteras conceptuales, la integración de evidencias heterogéneas y metodologías adaptativas para problemas complejos" (the contribution of artificial intelligence transcends operational efficiency, enabling new conceptual frontiers, integration of heterogeneous evidence, and adaptive methodologies for complex problems).

Research paradigm

transdisciplinary exploratory-analytical, combining hermeneutic-interpretive and critical technology studies

Author conclusions

The authors conclude that "La irrupción de la inteligencia artificial está remodelando estructuralmente la producción de conocimiento científico al convertirla en un espacio de cocreación entre agentes humanos y algorítmicos." They propose the "Interactive Theory of AI in Academic Knowledge Production" synthesized around five axes: AI as cognitive amplifier; human-AI interaction as recursive dialogue; human cognition as interpretive filter; shared authorship of knowledge artifacts; and ethical-epistemic validation oriented toward transparency, traceability, and accountability. The authors call for three action fronts: "i) investigación empírica comparativa sobre los efectos de la inteligencia artificial generativa en productividad, creatividad, reproducibilidad y diversidad disciplinar; ii) desarrollo normativo-institucional que regule ética, trazabilidad y acceso a infraestructuras de inteligencia artificial generativa, y iii) actualización curricular que dote a los investigadores de alfabetización algorítmica y competencias para la colaboración humano-máquina."

Risk of bias

Not explicitly stated in the paper.; Selection bias potential: document selection based on 'theoretical relevance and timeliness' may introduce researcher bias in corpus construction. No systematic review protocol or PRISMA compliance stated. No independent coding verification mentioned. Language bias possible (searches conducted in international databases but paper in Spanish suggests potential language-specific selection). No explicit criteria for inclusion/exclusion of sources documented.

Limitations

The authors acknowledge that "aún no existe un marco teórico general que pueda explicar completamente la contribución dinámica de la inteligencia artificial en la producción de conocimiento científico" (there still does not exist a general theoretical framework that can completely explain the dynamic contribution of artificial intelligence in the production of scientific knowledge)
Additionally, they note that "los modelos de las teorías del constructivismo y el conectivismo, y la cognición distribuida, proporcionan puntos de vista interesantes, pero carecen del poder explicativo para tener en cuenta la co-agencia, la adaptabilidad y la reflexividad" (the models of constructivism and connectivism theories, and distributed cognition, provide interesting viewpoints, but lack the explanatory power to account for co-agency, adaptability, and reflexivity).

Open questions raised

The authors identify: (1) Absence of a general theoretical framework explaining how generative AI contributes to scientific knowledge production; (2) Need for comparative empirical research on AI effects on productivity, creativity, reproducibility and disciplinary diversity; (3) Gap in normative-institutional development for regulating ethics, traceability and equitable access to AI infrastructure; (4) Need for curricular updates providing algorithmic literacy; (5) Requirement for rigorous epistemic governance addressing data bias, algorithmic opacity, authorship dilemmas and access gaps; (6) Necessity to develop robust ontologies and epistemologies specific to relational and generative science.
The authors identify the need for: (1) comparative empirical research on the effects of generative AI on productivity, creativity, reproducibility, and disciplinary diversity; (2) normative-institutional development to regulate ethics, traceability, and access to generative AI infrastructures; (3) curriculum updating to provide researchers with algorithmic literacy and competencies for human-machine collaboration. They emphasize the need for a comprehensive theoretical framework explaining how generative AI reshapes authorship, originality, legitimacy, and the nature of knowledge itself.
The authors identify several critical gaps: (1) Absence of a general theoretical framework explaining AI's dynamic contribution to scientific knowledge production; (2) Insufficient explanatory power of existing constructivist, connectivist, and distributed cognition models to account for co-agency, adaptability, and reflexivity in human-AI systems; (3) Need for ontology and epistemology specific to relational and generative science; (4) Persistent challenges including digital divide and unequal access to AI technologies; (5) Lack of robust ethical and epistemic validation modes ensuring reliability and legitimacy in hybrid knowledge generation.

Data: None mentionedCode: None mentionedExtracted from: pdfAgreement 74%

Explore related topics

Epistemic Risk Bias & Epistemic Risk Literature Review