AI evidence extraction

The Epistemic Authorship Crisis in the Age of Generative AI: Overcoming the Responsibility Gap through Hyper Justification Obligations

Rizky Fahmi Saputra, Mohammad Isa Wibisono, Agung Winarno, Subagyo Subagyo · International Journal of Economics, Commerce, and Management. · 2026

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

9/10

Relevance

0/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.62951/ijecm.v3i1.1090

Methodology & findings

Study design

Conceptual analysis approach involving three stages: (1) clarification of key concepts (epistemic luck, Gettier cases, hallucination, justification forms), (2) comparative analysis between epistemic structure in Gettier cases and probabilistic mechanisms of LLMs to identify analogous patterns formulated as Algorithmic Gettier Cases (AGCs), and (3) normative analysis of ethical consequences of LLM use on scientific authorship accountability.

Main result

The study found that "Algorithmic Gettier Cases (AGCs) occur when linguistic coherence deceives users and creates the impression of justification, even though the truth that emerges is statistical coincidence and is not supported by valid causal relationships." The research demonstrates that "LLMs can generate text that appears justified and correct, but is often based on epistemic luck (accidental truth) without a valid causal relationship" and that "AGCs are a systemic phenomenon arising from the probabilistic architecture of LLMs, unlike classical Gettier cases which are local in nature."

Research paradigm

Philosophical argumentation and conceptual analysis

Author conclusions

"This article demonstrates that the use of Large Language Models (LLMs) in scientific knowledge production creates a new form of epistemic luck that is structurally analogous to Gettier cases." The authors conclude that "the main contribution of this research is to develop a conceptual framework that can help the academic community understand the epistemic risks of LLM and provide a normative foundation that can strengthen the integrity of science in the era of generative AI." They propose that the "Hyper-Justification Obligation as an ethical principle that researchers need to adopt when using AI in scientific writing" requires "active verification, causal tracing, and reaffirmation of the role of humans as the ultimate holders of epistemic authority."

Risk of bias

No empirical validation of AGC concept in actual research publications; Reliance on cited studies without direct computational testing; Lack of quantitative frequency measurements of AGCs; No longitudinal studies measuring AGC occurrence rates; No systematic bias assessment is provided. As a conceptual/philosophical analysis rather than empirical research, traditional bias risks (selection bias, attrition, confounders) do not apply. However, the analysis relies on cited empirical studies which may have their own biases.

Limitations

The authors acknowledge that "belum ada studi longitudinal kuantitatif yang secara langsung mengukur frekuensi AGCs dalam korpus publikasi akademik" (there are no longitudinal quantitative studies that directly measure the frequency of AGCs in academic publication corpora), and note that "human verification procedures are effective on a case-by-case basis but are costly and not scalable, leaving risks inherent at the system level." Additionally, empirical evidence presented is conceptual in nature rather than based on direct empirical testing of AGC phenomena in actual research contexts.

Open questions raised

The literature gap addressed includes the absence of in-depth studies explaining how the probabilistic nature of LLMs affects the structure of scientific accountability and authorship status. The authors note that although studies have highlighted hallucinations and epistemic limitations of AI, there has been no analysis linking the risk of epistemic luck in AI outputs to the moral and normative responsibilities of human authors.
The authors identify the absence of in-depth studies explaining how the probabilistic nature of LLMs affects the structure of scientific accountability and authorship status. They note that although studies have highlighted hallucinations and epistemic limitations of AI, there has been no analysis linking the risk of epistemic luck in AI outputs to the moral and normative responsibilities of human authors.
The authors identify that "the literature gap addressed in this article lies in the absence of in-depth studies explaining how the probabilistic nature of LLMs affects the structure of scientific accountability and authorship status." They note that "although a number of studies have highlighted the hallucinations and epistemic limitations of AI, so far there has been no analysis linking the risk of epistemic luck in AI outputs to the moral and normative responsibilities of human authors." Future research directions include the need for quantitative longitudinal studies measuring AGC frequency in academic publication corpora.

Extracted from: pdfAgreement 79%

Explore related topics

Epistemic Risk Research Integrity Conceptual