AI evidence extraction

Defining and assessing AI literacy for researchers across the research lifecycle

Jessica L. Parker, Kimberly P. Becker · Frontiers in Education · 2026

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/10.3389/feduc.2026.1827603

Methodology & findings

Study design

Qualitative mixed-methods approach combining: (1) conceptual/theoretical framework development grounded in Bourdieu's cultural intermediary concept and Selber's digital literacy model; (2) practitioner engagement through workshops and webinars delivered to several hundred researchers from 2023-2025 (non-systematically recorded); (3) iterative refinement informed by prior published research on researcher-AI interactions (Parker et al., 2023a, 2023b, 2023c, 2025); (4) development of a capability map organizing AI literacy across research lifecycle stages and three literacy domains; (5) creation of a performance-based assessment rubric with three performance levels (emerging, developing, proficient)..

Main result

The paper establishes that "AI functions as a cultural intermediary in research" by "filter[ing], rank[ing], summariz[ing], and model[ing] scholarly expression through pattern reproduction derived from training data instead of through deliberate curation." The framework defines AI literacy as comprising three interconnected dimensions—functional, critical, and rhetorical literacies—mapped across the research lifecycle (research question formulation, literature review, methods selection, data analysis, writing, peer review, and dissemination). The authors argue that "researchers need to recognize when AI is shaping scholarly norms, evaluate what that means for their work, and step in deliberately rather than accepting AI output as neutral or authoritative."

Research paradigm

Interpretive/Critical theory (Bourdieu-influenced socio-cultural analysis)

Author conclusions

The authors conclude that "AI functions as a cultural intermediary in research" and that "Framing AI literacy as cultural mediation shifts attention back to scholarly practice. It affirms that while AI may participate in research, responsibility for knowledge production remains human." They assert that "By integrating functional competence, critical judgment, and rhetorical agency, the framework provides a foundation for institutions to support innovation while safeguarding research integrity and public trust." The central argument is that "the central challenge is not how to control these systems, but how to support researchers in working with them responsibly."

Risk of bias

Selection bias in practitioner engagement: workshops and webinars were not systematically designed for data collection; feedback was not formally recorded; Lack of formal validation of the framework across diverse disciplinary contexts; Framework reflects insights from researchers who voluntarily attended AI literacy sessions (potential selection bias toward more engaged researchers); Discipline-agnostic framing may obscure field-specific norms and practices; Selection bias in practitioner engagement: workshops and webinars were self-selected researchers, not representative sample; Feedback not systematically recorded: reliance on authors' recollection of 'recurring questions' rather than structured data collection; Lack of disciplinary representation: framework developed primarily from engagement across institutional/disciplinary contexts without systematic sampling; Author positionality: authors co-founded an AI-native educational technology company serving researchers, potential financial/professional interest in AI adoption; Lack of formal validation: no independent assessor reliability testing; Selection bias in practitioner engagement: workshop participants may not be representative of all researchers; Lack of systematic data collection from practitioner sessions—feedback was not systematically recorded; Framework development informed primarily by English-language research contexts and institutional settings; Potential bias toward Western, industrialized research practices in framework design

Limitations

The authors acknowledge that "the capability map and rubric presented here represent a theoretically grounded and practitioner-informed first generation of tools rather than a validated instrument." They note that "formal validation has not yet been conducted" and that "the framework's discipline-agnostic design is also a limitation worth naming" because "the relative weighting of functional, critical, and rhetorical literacies is likely to vary across fields, and disciplinary adaptation may be necessary for applied contexts such as clinical research or computational sciences." Additionally, "the framework does not account for the fact that AI systems themselves will continue to change in ways that outpace any static capability model."

Open questions raised

Formal validation of the capability map and rubric across assessors and disciplinary contexts
Longitudinal research examining whether engagement with the framework leads to measurable changes in researcher documentation and reflection on AI use
Systematic disciplinary adaptation of the framework for field-specific norms and practices
Examination of the framework's applicability to emerging AI capabilities including multimodal generation, autonomous research agents, and real-time collaboration interfaces
Expert panel review involving research integrity officers, doctoral supervisors, and early-career researchers to evaluate face and content validity
Formal validation of the capability map and rubric through supervisor application and assessor reliability testing

Extracted from: pdfAgreement 73%

Explore related topics

AI Literacy Training Doctoral Training Conceptual