AI evidence extraction

AI Dependency vs. Doctoral Identity: How Generative AI is Challenging the Development of Independent Scholarly Thinking in Doctoral Students

Valerie A. Storey · International Journal of AI in Pedagogy Innovation and Learning Futures · 2026

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

9/10

Relevance

1/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.46787/ijaipil.v2026i1.6971

Methodology & findings

Study design

Critical literature review across four major databases (ERIC, Scopus, Web of Science, Google Scholar) supplemented by federal legislation, professional organization working papers, and reports published between late 2022 and March 2026.

Sample

not–applicable–to–review

Primary method

Inductive synthesis of heterogeneous literature. No quantitative meta-analysis or pooled statistics conducted. The review uses qualitative thematic synthesis and conceptual mapping (see Table 1: Literature-to-Framework Mapping) to organize sources by dimension of risk (Critical Thinking, Knowledge Authenticity, Doctoral Identity).

Main result

The paper identifies three interconnected dimensions of risk in unguided doctoral GenAI adoption: "(1) critical thinking attrition, (2) knowledge authenticity erosion, and (3) doctoral identity disruption." The authors argue that "A doctoral student who offloads analytical judgment to a GenAI is simultaneously undermining their critical thinking capacity, producing work whose authenticity cannot be verified, and bypassing the developmental struggle through which a scholarly identity is formed." Additionally, research shows that "63% of EdD doctoral students in the U.S. report active use of GenAI for proofreading (49%), academic writing (41%), idea generation (39%), and literature reviews (33%)" yet "only 36% of students report receiving formal AI skills training from their institution."

Reports effect sizes.

Research paradigm

Critical interpretivism

Author conclusions

The authors conclude that "The tools that enhance doctoral research productivity may, when adopted without institutional guidance, also undermine the development of the scholarly independence that doctoral education exists to produce." They argue that "The conceptual distinction between GenAI as a productivity tool, and GenAI as a scholarly surrogate does not currently exist in institutional policy at any tier. This is the condition under which all three dimensions of risk arise and compound one another." The authors propose that "Doctoral programs need to develop competency-based frameworks that explicitly operationalize where legitimate GenAI use ends and surrogacy begins, differentiated by each stage of the research process" and that "If the answer is to remain the doctoral student, and the doctorate continues to certify that an individual has demonstrated the capacity for original, independent scholarly contribution, then the productivity-surrogate distinction must become the foundation of doctoral AI policy."

Risk of bias

Selection bias in literature inclusion (sources retained based on relevance to review question); Interpretation bias inherent to inductive synthesis methodology; Temporal bias: search window begins after ChatGPT release (late 2022), potentially missing pre-GenAI foundational work; Potential reporting bias in reviewed studies on doctoral GenAI use; No independent screening or blinded review process reported; Selection bias in literature search: limited to sources in English and major academic databases; Publication bias: review relies on published sources which may skew toward negative findings; Interpretive bias: inductive synthesis involves author judgment in theme identification; Geographic bias: disproportionate focus on U.S., Canadian, and European institutions; Temporal bias: rapid field development means some included sources may be quickly superseded; Grey literature bias: inclusion of federal reports and professional organization papers alongside peer-reviewed research; Confounding factors: inability to isolate GenAI effects from institutional governance quality, faculty training, and student preparation; Interpretive judgment in inductive synthesis may introduce researcher bias; Rapidly evolving field may limit generalizability of findings; Limited empirical evidence base on doctoral-specific outcomes; Heterogeneous body of evidence may include studies of varying quality

Limitations

The paper acknowledges that "the field is recent and the evidence base is still developing" and notes that "the identification of the three dimensions reflects the interpretive judgment that the critical review method both permits and requires." The authors state the work is "offered as the paper's original conceptual contribution rather than as a claim of definitive causal proof." Additionally, the paper notes that "the impact of GenAI on critical thinking remains underexplored" and identifies that "future research should examine whether doctoral students who engage in high levels of unguided GenAI demonstrate measurable differences in independent scholarly capacity at career entry."

Open questions raised

Doctoral-specific outcomes research: "Future research should examine whether doctoral students who engage in high levels of unguided GenAI demonstrate measurable differences in independent scholarly capacity at career entry, including in publication quality, peer review judgements, and research leadership. The evidence base on doctoral specific outcomes remains thin and this gap is a research priority."
Absence of institutional frameworks distinguishing productivity from surrogacy at doctoral level
Limited research on critical thinking impacts in 2026: "In 2026, the impact of GenAI on critical thinking remains underexplored"
Inadequate guidance on original contribution in AI-mediated environments
Need for process-level documentation and staged disclosure protocols for dissertations
Doctoral-specific outcomes research: "Future research should examine whether doctoral students who engage in high levels of unguided GenAI demonstrate measurable differences in independent scholarly capacity at career entry, including in publication quality, peer review judgements, and research leadership."

Extracted from: pdfAgreement 66%

Explore related topics

AI Literacy Training Doctoral Training Literature Review