AI evidence extraction

Probabilistic Obliteration and Formulaic Fabrication

Joel Blechinger · KULA knowledge creation dissemination and preservation studies · 2026

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

9/10

Relevance

3/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.18357/kula.310

Methodology & findings

Study design

Hermeneutic textual analysis and critical discourse analysis examining two sets of examples: (1) official style guide citation guidance for GenAI tools (MLA, APA, CMOS), and (2) fabricated citations generated by GenAI systems.

Main result

The study found that "GenAI is antithetical to citational justice" and that "LLMs engage in a kind of dynamic and probabilistic obliteration effect" where "present technological obliteration—as I have argued elsewhere—liquidates authorship and attribution, stripping authors of even the paltry symbolic remuneration accorded them through citational recognition." The author argues that citation guidance from style guides like the APA reifies corporate enclosure by naming the technology company as author, and that fabricated citations reveal citation has become "a formulaic rhetorical gesture that can be imitated in formal terms but is fundamentally evacuated of meaning."

Reports effect sizes.

Research paradigm

Critical/Interpretive

Author conclusions

The author concludes that "to attempt to speak of citational justice—let alone material justice—in light of such profoundly irrelational technologies leaves one uncertain where to even begin." The author proposes five reflective questions for reimagining library instruction: "How can we take the opportunity that GenAI presents for our pedagogy to radically reimagine library instruction around the importance of attribution and relationality in academic work?" and "What would it mean to teach against the grain of hegemonic style guides like the APA, and to disagree productively—emphatically and explicitly—with their official guidance in our own pedagogy?" The author concludes: "Of course, these avenues for pedagogical exploration of GenAI citation issues would not singlehandedly bring about citational justice in the GenAI context, but they would at least clear space for us to begin to think through it again in the wake of increased use of these fundamentally irrelational technologies."

Risk of bias

As a theoretical/argumentative paper rather than empirical research, traditional bias categories (selection bias, attrition bias) do not apply. However, potential limitations include: author's stated positionality as library professional may influence interpretation; analysis focuses primarily on style guide examples rather than systematic sampling; no systematic search methodology described for identifying GenAI fabrication examples; reliance on pre-2024 models and guidance that have been updated during writing.; Author's explicit ideological commitment to citational justice movement may shape interpretation; Selection of examples may be non-representative; Temporal limitation: analysis based on early GenAI models (ChatGPT, Gemini) without RAG; No systematic search strategy for fabrication examples

Limitations

The author notes that "the perpetual constraints of the standard one-shot instructional session (Nicholson 2016
Pagowsky 2021
Almeida 2022) may hamper one's ability to integrate citational justice content as thoroughly as desired." Additionally, the author acknowledges that "Admittedly, much of my thinking in this article related to GenAI fabricated citations was developed when earlier general-purpose models without RAG functionality, like ChatGPT and Gemini, dominated the GenAI conversation," suggesting the analysis may not fully address more recent retrieval-augmented generation approaches.

Open questions raised

Further engagement with RAG-specific citational justice issues could extend this work in a subsequent article
Deeper exploration of how to integrate citational justice content into library instruction despite constraints
Investigation of how legitimate diversity in GenAI training data compounds rather than solves citational justice issues
Critical analysis of how to teach citation as relational practice rather than from compliance and fear
The author identifies the need for: (1) deeper engagement with citational justice issues in the context of retrieval-augmented generation (RAG) systems; (2) radical reimagining of library instruction around attribution and relationality rather than compliance; (3) integration of material harms (climate impact, exploitative labor) with analyses of citational harms; (4) pedagogies that push against hegemonic style guides; (5) future work extending analysis of RAG-specific citational justice issues in a subsequent article.
The author identifies several gaps: (1) insufficient critical engagement with what it means to cite GenAI tools beyond mechanical citation style adjustments; (2) limited analysis of how GenAI citation relates to broader neoliberal transformation of higher education; (3) need for RAG-specific citational justice analysis (noted as potential future work); (4) lack of integration of citational justice content in library instruction due to "perpetual constraints of the standard one-shot instructional session"; (5) insufficient attention to materiality of GenAI harms (climate impact, exploitative labor practices) in citation scholarship.

Code: cleanBib code base; One workshop slide repository mentioned: https://tinyurl.com/y9nnkmd4Extracted from: pdfAgreement 75%

Explore related topics

Citation Integrity Scholarly Infrastructure Conceptual