CoChat
Sign Up
Cited by 0Artificial Intelligence in EducationEducational Innovations and ChallengesArtificial Intelligence in Healthcare and Education

A Multi-Dimensional Decision Framework for Evaluating and Selecting AI-Enabled Instructional Technology in U.S. Higher Education

Open access
Akter, Fatema, Parvatha Reddy, Kiran Kumar·2025·International Journal For Multidisciplinary Research
DOI 10.36948/ijfmr.2025.v07i06.84921
Open this paper in CoChat — free accountOpen PDF

Abstract

We present a multi-dimensional decision framework for evaluating and selecting AI-enabled instructional technology in U.S. higher education institutions. The research addresses the critical gap between rapid generative AI advancement and the need for rigorous, policy-aligned institutional procurement processes. Our methodology employed a multi-phase systematic literature review and meta-analysis covering publications from 2013 to 2026, sourced from the Scopus, Web of Science, ERIC, and IEEE Xplore databases. Phase one synthesized empirical studies on adaptive learning, intelligent tutoring systems, and generative AI tools. The second phase employed the Unified Theory of Acceptance and Use of Technology (UTAUT) framework to conduct technology acceptance research, whereas the third phase examined institutional procurement records and vendor assertions. The findings indicate that performance expectancy and effort expectancy are the foremost predictors of behavioral intention among faculty and students to adopt AI tools. Prompt engineering strategies yield measurable improvements in feedback quality and workflow efficiency; for instance, one study noted a 34.6% reduction in grading time due to iterative prompts. However, effectiveness remains highly sensitive to specific tasks and evaluation criteria complexity. Institutional procurement analysis identifies FERPA compliance, WCAG 2.1 AA accessibility standards, and data residency requirements as non-negotiable criteria. A critical oversight in governance continues with respect to personal accounts, as faculty and students completely circumvent institutional FERPA protections. The proposed framework integrates pedagogical effectiveness, technology acceptance factors, prompt engineering strategies, and institutional governance metrics into a coherent decision tool. This work presents a methodical, empirically grounded framework linking technological progress with institutional responsibility, thereby empowering higher education administrators to reach well-informed, policy-conforming choices regarding the adoption of AI tools while reducing hazards associated with academic honesty, data confidentiality, and fairness.

AI snapshot

Written from the paper’s full text.

TL;DR

This paper proposes a multi-dimensional decision framework for selecting AI-enabled instructional technology in U.S. higher education, synthesizing evidence on pedagogical effectiveness, UTAUT acceptance, prompt engineering, and procurement governance. It finds performance and effort expectancy as top adoption predictors, prompt engineering can improve efficiency but is task-limited, and FERPA/accessibility/data-residency compliance are non-negotiable while personal AI accounts create a major governance gap.

Key findings

  • UTAUT review: performance expectancy was the strongest and most consistent predictor of behavioral intention, effort expectancy was second, and trust, privacy concerns, and perceived ethical risk emerged as additional predictors beyond the standard UTAUT framework.
  • Prompt engineering: explicit, iterative, and task-aligned prompts improved feedback quality and workflow efficiency; one grading-support study reported a 34.6% reduction in grading time, but modifying grading instructions did not substantially alter GPT-4’s holistic essay evaluation.
  • Procurement/governance: FERPA compliance, WCAG 2.1 AA accessibility, and data residency were non-negotiable criteria; vendors were required to be designated a “school official,” with TLS 1.2+ in transit, AES-256 at rest, audit rights, VPAT, and breach notification commonly within 72 hours.
  • Evidence gaps: adaptive learning systems such as Carnegie Learning’s MATHia had robust independent evidence with up to 36% higher knowledge retention than traditional instruction, while generative AI vendor claims often lacked independent empirical support; personal AI account use by faculty and students bypassed institutional FERPA protections.

Methods

  • Three-phase systematic literature review and meta-analysis: Phase 1 searched Scopus, Web of Science, ERIC, and IEEE Xplore for peer-reviewed journal articles, conference proceedings, and book chapters published January 2013–December 2026; extracted effect sizes and used a random-effects meta-analysis with I², Cochran’s Q, funnel plot asymmetry, and Egger’s regression test.
  • Phase 2 targeted generative AI/LLM studies from 2016–2026 and UTAUT/extension studies, extracting prompt engineering strategies and effect sizes for constructs predicting behavioral intention.
  • Phase 3 analyzed institutional procurement policies, vendor documentation, and guidance documents; classified vendor evidence as strong, moderate, weak, or absent and used convergent mixed-methods synthesis across phases.

Limitations

  • Stated: the search was limited to four databases and English-language publications; rapid AI change may make older findings obsolete; generative AI literature relies on short-term interventions; quality appraisal involved subjective judgments; publication bias may overestimate effectiveness; heterogeneity limited quantitative synthesis.
  • Evident from the text: despite describing a meta-analysis, the paper reports few pooled meta-analytic effect sizes or heterogeneity statistics, and several claims rely on cited secondary reviews rather than newly conducted primary data collection.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.36948/ijfmr.2025.v07i06.84921

Cites

21 of 21 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

15 of 34 cited sources checked · 1 supported · 0 flagged · accuracy not yet measured

  • Findingcitation

    One study reported a 34.6% reduction in grading time due to iterative prompts.

    “one study noted a 34.6% reduction in grading time due to iterative prompts.”
    • Also stated in: § Outcomes and Challenges of AI-Enabled Instructional Technology · plain
      “One study on grading support reported a 34.6% reduction in grading time via iterative prompt engineering, which constitutes a substantial efficiency gain for instructors (Wan and Chen, 2024).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Wan and Chen, 2024 source

      The Abstract (p. 1) says the researchers refined their prompt over several iterations, and the Discussion (p. 14) suggests the approach could save grading time. The paper does not report a measured 34.6% reduction in grading time; its instructor ratings assessed how much editing the generated feedback would need (Stage 3, p. 8). (LLM, conf=0.86)

      Experimental — accuracy not yet measured
  • Findingcitation

    WCAG 2.1 AA accessibility standards are a non-negotiable criterion in institutional procurement analysis.

    “Institutional procurement analysis identifies FERPA compliance, WCAG 2.1 AA accessibility standards, and data residency requirements as non-negotiable criteria.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Stelea, Robu and Sandu, 2025 source

      Section 2.1 discusses WCAG 2.1 compliance in e-learning platforms, and Section 3.2.1 says the platform follows WCAG 2.1 guidelines. Neither passage addresses WCAG 2.1 AA as a non-negotiable institutional procurement criterion; searches for procurement, FERPA, data residency, and WCAG 2.1 AA did not locate such an analysis in our copy. (LLM, conf=0.74)

      Experimental — accuracy not yet measured
  • Premisecitation

    Recent expansions of UTAUT have integrated trust, privacy concerns, and ethical considerations as critical moderators.

    “Recent expansions of UTAUT have integrated trust, privacy concerns, and ethical considerations as critical moderators, which mirrors the distinctive challenges introduced by AI systems (Rana et al., 2024).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Rana et al., 2024 source

      The Discussion’s Theoretical contributions section says the study extends UTAUT by integrating trust and privacy as key constructs for understanding AI adoption. It does not identify ethical considerations as an integrated construct or describe these factors as moderators; the paper discusses trust and privacy chiefly in relation to adoption intentions and use behavior. Page numbers are unavailable in our copy. (LLM, conf=0.90)

      Experimental — accuracy not yet measured
  • Findingcitation

    AI systems trained on historical data can perpetuate and amplify existing biases related to race, gender, socioeconomic status, and other demographic factors.

    “AI systems trained on historical data can perpetuate and amplify existing biases related to race, gender, socioeconomic status, and other demographic factors (Anyaso and Anyaso, 2025).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Anyaso and Anyaso, 2025 source

      Section 4.4, “Bias in AI Algorithms,” says that algorithms trained on data reflecting existing biases may perpetuate or amplify those biases and produce discriminatory outcomes for some students. The passage does not identify race, gender, socioeconomic status, or other specific demographic factors, and frames the concern in higher education. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
  • Findingcitation

    Multiple studies documented cases where students' ability to perform tasks independently declined after extended application of AI tools.

    “Multiple studies documented cases where students' ability to perform tasks independently declined after extended application of AI tools, which raised concerns about the long-term impact on critical thinking, composition, and problem-solving skills (Ebrahim, Mhouti and Allaoui, 2025).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Ebrahim, Mhouti and Allaoui, 2025 source

      Section 4.2 discusses concerns about assessment credibility and academic integrity, while section 4.1 discusses AI-assisted task completion and students’ autonomy to learn (page numbers unavailable). Neither section reports studies documenting a decline in students’ ability to perform tasks independently after extended AI use. Searches for independent task performance, dependency, extended use, and long-term effects found no passage establishing that claim. (LLM, conf=0.74)

      Experimental — accuracy not yet measured
  • Classified as probably not a substantial claim. 11
  • May combine two claims. 26
  • The quote only partly states this claim. 2

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The quality and composition of prompts are the main determinants of LLM output quality in educational contexts.

    “Research indicates that the quality and composition of prompts are the main determinants of LLM output quality in educational contexts (Sahoo et al., 2024).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Sahoo et al., 2024 source

      The Introduction (page not provided) says carefully crafted prompts can steer and improve LLM outputs, and the Few-Shot Prompting section says the selection and composition of examples can significantly influence model behavior. The cited passages do not establish that prompt quality and composition are the main determinants of output quality, particularly in educational contexts. (LLM, conf=0.87)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingcitation

    The rapid emergence of generative AI and large language models after 2022 fundamentally shifted the research landscape, introducing both new opportunities and unprecedented challenges.

    “However, the rapid emergence of generative AI and large language models after 2022 fundamentally shifted the research landscape, introducing both new opportunities and unprecedented challenges (Ilieva et al., 2026).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Ilieva et al., 2026 source

      The Introduction (page not provided) says recent generative AI is rapidly transforming higher education, creating new capabilities alongside technical and ethical challenges. It does not establish a shift in the broader research landscape specifically after 2022, or characterize the challenges as unprecedented. (LLM, conf=0.83)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingcitation

    Adherence to WCAG 2.1 AA and Section 508, along with the provision of a Voluntary Product Accessibility Template (VPAT), is mandatory.

    “Adherence to WCAG 2.1 AA and Section 508, along with the provision of a Voluntary Product Accessibility Template (VPAT), is mandatory (Stelea, Robu and Sandu, 2025).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Stelea, Robu and Sandu, 2025 source

      Section 2.4, “Identified Gaps” (page unavailable), calls Section 508 a legal mandate and refers to W3C guidelines. The paper also discusses WCAG 2.1 in its front-end design section, but the passages do not establish that WCAG 2.1 AA compliance or provision of a VPAT is mandatory. (LLM, conf=0.85)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingcitation

    Academic integrity issues constitute the most immediate and broadly debated topic, as research has shown that students extensively employ generative AI tools to finish assignments without permission, generate plagiarized work, and circumvent conventional assessment methods.

    “Academic integrity issues constitute the most immediate and broadly debated topic, as research has shown that students extensively employ generative AI tools to finish assignments without permission, generate plagiarized work, and circumvent conventional assessment methods (Ebrahim, Mhouti and Allaoui, 2025).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Ebrahim, Mhouti and Allaoui, 2025 source

      Split into 2 atoms — 0 confirmed, 2 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingcitation

    Students employing AI tools for writing assistance achieved drafts with superior organization, grammar, and argumentation compared to control groups, and these gains were retained in later revisions.

    “In comparison to control groups, students employing AI tools for writing assistance achieved drafts with superior organization, grammar, and argumentation, and these gains were retained in later revisions (Sahoo et al., 2024).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Sahoo et al., 2024 source

      The Self-Refine Prompting section discusses LLMs revising their own outputs, not students using AI writing assistance. Searches for student writing, grammar, control groups, and later revisions found no passage in our copy reporting the claimed comparison or retained gains. The cited paper’s passages therefore do not establish this claim. (LLM, conf=0.76)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingcitation

    Following 2022, research on generative AI and large language models increased markedly, and ChatGPT was the tool most frequently examined in every application category.

    “Nevertheless, the period following 2022 saw a marked increase in research on generative AI and large language models, and ChatGPT was the tool most frequently examined in every application category (Ilieva et al., 2026).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Ilieva et al., 2026 source

      The Introduction discusses recent generative AI in higher education, but does not report a post-2022 increase in research. Searches for the post-2022 trend and for ChatGPT found no passage comparing how frequently tools were examined across application categories; the paper’s empirical application evaluates courses instead. (LLM, conf=0.76)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingcitation

    Many institutions have not yet developed comprehensive policies governing AI tool usage, so faculty and students must navigate an uncertain landscape without clear guidance.

    “Many institutions have not yet developed comprehensive policies governing AI tool usage, so faculty and students must navigate an uncertain landscape without clear guidance (Teleki et al., 2026).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Teleki et al., 2026 source

  • May combine two claims.

    Findingcitation

    Iterative, task-aligned prompts yield measurable gains in feedback quality, workflow efficiency, and student engagement.

    “Iterative, task-aligned prompts yield measurable gains in feedback quality, workflow efficiency, and student engagement.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingcitation

    Prompt engineering strategies yield measurable improvements in feedback quality and workflow efficiency.

    “Prompt engineering strategies yield measurable improvements in feedback quality and workflow efficiency; for instance, one study noted a 34.6% reduction in grading time due to iterative prompts.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given

    Assessment and academic integrity have emerged as dominant thematic areas in the post-ChatGPT era.

    “Assessment and academic integrity have emerged as dominant thematic areas in the post-ChatGPT era.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingcitation

    Institutions are increasingly requiring explicit adherence to data privacy and security standards, with FERPA compliance being the most important criterion in procurement decisions.

    “The results suggest that institutions are increasingly requiring explicit adherence to data privacy and security standards, with FERPA compliance being the most important criterion in procurement decisions (Purdy, 2025).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Purdy, 2025 source

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Findingnone given

    The evidence base for learning analytics dashboards remains nascent, and investigations have documented inconsistent effects on student achievement contingent on dashboard design, user training, and degree of alignment with pedagogical approaches.

    “The evidence base for learning analytics dashboards remains nascent, and investigations have documented inconsistent effects on student achievement that are contingent on dashboard design, user training, and the degree of alignment with pedagogical approaches.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingcitation

    Adaptive learning systems constituted the most developed category, possessing a thoroughly established evidence base spanning multiple disciplines and educational levels.

    “Adaptive learning systems constituted the most developed category, possessing a thoroughly established evidence base that spans multiple disciplines and educational levels.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    The evidence base for generative AI tools is defined by short-term intervention studies and a predominance of qualitative or quasi-experimental designs.

    “The evidence base for generative AI tools is defined by short-term intervention studies and a predominance of qualitative or quasi-experimental designs, which is a consequence of the recent nature of their widespread adoption.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingcitation

    For adaptive learning systems, the evidence base was generally strong, as multiple independent studies validated vendor claims of improved learning outcomes.

    “For adaptive learning systems, the evidence base was generally strong, as multiple independent studies validated vendor claims of improved learning outcomes (Kulik and Fletcher, 2016).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Kulik and Fletcher, 2016 source

  • May combine two claims.

    Findingcitation

    Studies recorded cases in which student data was transmitted to third-party AI providers without sufficient consent or data use agreements, which may contravene federal regulations.

    “Studies recorded cases in which student data was transmitted to third-party AI providers without sufficient consent or data use agreements, which may contravene federal regulations (Purdy, 2025).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Purdy, 2025 source

  • May combine two claims.

    Findingcitation

    The evidence base for chatbots exhibits greater heterogeneity than that for adaptive learning systems, and the effectiveness of chatbots depends substantially on their design, deployment context, and interaction type.

    “The evidence base for chatbots exhibits greater heterogeneity than that for adaptive learning systems, and the effectiveness of chatbots depends substantially on their design, deployment context, and interaction type (Zhao, 2024).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Zhao, 2024

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Findingreasoning

    Explicit, iterative, and task-aligned prompts produce measurable improvements in feedback quality, workflow efficiency, student engagement, and prompt literacy.

    “The evidence consistently showed that explicit, iterative, and task-aligned prompts produced measurable improvements in multiple aspects of educational practice, such as feedback quality, workflow efficiency, student engagement, and the cultivation of prompt literacy among users.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingcitation

    AI-enabled instructional technology supports improved idea generation, critical thinking, and problem-solving skills.

    “Regarding cognitive outcomes, the evidence suggests that AI-enabled instructional technology supports improved idea generation, critical thinking, and problem-solving skills.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingcitation

    Prompt literacy is trainable and beneficial for both faculty and students.

    “Prompt literacy, defined as the ability to design, evaluate, and refine prompts for specific educational purposes, was found to be trainable and beneficial for both faculty and students.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given

    Suppliers must be classified as a school official under a strict constraint on scope that prohibits the employment of student data for model training absent consent.

    “Suppliers must be classified as a “school official” under a strict constraint on scope that prohibits the employment of student data for model training absent consent.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Findingnone given

    Studies on generative AI tools tend to have shorter intervention periods, smaller sample sizes, and less rigorous control conditions compared to the established literature on adaptive learning systems.

    “studies on generative AI tools tend to have shorter intervention periods, smaller sample sizes, and less rigorous control conditions compared to the established literature on adaptive learning systems.”
    Quote in the paperNot a substantial claim

  • May combine two claims. The quote only partly states this claim.

    Findingcitation

    Accessibility compliance emerged as a non-negotiable criterion in institutional procurement processes, with WCAG 2.1 AA as a baseline.

    “Accessibility compliance emerged as a non-negotiable criterion in institutional procurement processes.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingcitation

    For generative AI tools, the evidence base was weak, and most vendor claims were not supported by independent empirical evidence.

    “For generative AI tools, the evidence base was weak, and most vendor claims were not supported by independent empirical evidence”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Williamson and Eynon, 2020 source

  • May combine two claims.

    Premisecitation

    Effort expectancy is the second most impactful factor, whereas social influence and facilitating conditions fulfill more context-dependent roles.

    “Effort expectancy, also termed perceived ease of use, is the second most impactful factor, whereas social influence and facilitating conditions fulfill more contextdependent roles (Dwivedi et al., 2011).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Dwivedi et al., 2011 source

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    The proposed framework integrates pedagogical effectiveness, technology acceptance factors, prompt engineering strategies, and institutional governance metrics into a coherent decision tool.

    “The proposed framework integrates pedagogical effectiveness, technology acceptance factors, prompt engineering strategies, and institutional governance metrics into a coherent decision tool.”
    • Also stated in: § Literature Review · plain
      “The proposed framework addresses these limitations by integrating pedagogical effectiveness, technology acceptance factors, prompt engineering strategies, and institutional governance metrics into a coherent decision tool”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    UTAUT has been extensively validated in educational settings, and performance expectancy has consistently been found to be the strongest predictor of behavioral intention.

    “The Unified Theory of Acceptance and Use of Technology (UTAUT) has been extensively validated in educational settings, and performance expectancy has consistently been found to be the strongest predictor of behavioral intention (Venkatesh et al., 2003).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Venkatesh et al., 2003 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Modifying grading instructions through prompts did not substantially alter GPT-4's performance on essay evaluation.

    “some research has indicated that modifying grading instructions through prompts did not substantially alter GPT-4's performance on essay evaluation (Liew and Tan, 2024).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Liew and Tan, 2024 source

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    The study employed a multi-phase systematic literature review and meta-analysis covering publications from 2013 to 2026 across Scopus, Web of Science, ERIC, and IEEE Xplore.

    “Our methodology employed a multi-phase systematic literature review and meta-analysis covering publications from 2013 to 2026, sourced from the Scopus, Web of Science, ERIC, and IEEE Xplore databases.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    Prompt engineering does not serve as a universal solution for every shortcoming of large language models in educational settings; its efficacy is constrained by the inherent capacities and constraints of the model itself.

    “This discovery suggests that prompt engineering does not serve as a universal solution for every shortcoming of large language models within educational settings; instead, its efficacy is constrained by the inherent capacities and constraints of the model itself.”
    Quote in the paperSubstantial claim

  • The quote only partly states this claim.

    Interpretationcitation

    Social influence's effect on adoption intentions is context-dependent, varying with departmental attitudes toward AI.

    “This contextual dependence implies that institutional approaches to promoting AI adoption must be customized to disciplinary norms”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    The opacity of many AI systems hinders the detection and correction of biases, raising worries about equity and fairness in AI-mediated education.

    “The fact that many AI systems are opaque hinders the detection and correction of these biases, which sparks worries about equity and fairness in education mediated by AI.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Many AI systems are opaque.

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Interpretationreasoning

    This governance gap constitutes a considerable institutional vulnerability, because institutions can be held accountable for FERPA violations even if those violations arise from faculty or students' unauthorized or unmonitored use of personal accounts.

    “This governance gap constitutes a considerable institutional vulnerability, because institutions can be held accountable for FERPA violations even if those violations arise from faculty or students' unauthorized or unmonitored use of personal accounts.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Institutions can be held accountable for FERPA violations arising from faculty or students' unauthorized or unmonitored use of personal AI accounts.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    Publication bias may have led to an overestimation of the effectiveness of AI-enabled instructional technology.

    “Publication bias, where studies with positive findings are more likely to be published than those with null or negative results, may have led to an overestimation of the effectiveness of AI-enabled instructional technology.”
    Quote in the paperNot a substantial claim
34 of 34

  • The paper relies on this without support.

    Many AI systems are opaque.

    “The fact that many AI systems are opaque hinders the detection and correction of these biases, which sparks worries about equity and fairness in education mediated by AI.”

  • The paper relies on this without support.

    Detection of AI-generated content remains unreliable.

    “Detection of AI-generated content remains unreliable, and the cat-and-mouse dynamic between detection tools and increasingly sophisticated LLMs creates ongoing uncertainty for institutions.”

  • The paper relies on this without support.

    The cost of premium AI tools may create a two-tiered system where students with financial resources have access to more powerful AI support than their peers.

    “Furthermore, the cost of premium AI tools may create a twotiered system where students with financial resources have access to more powerful AI support than their peers.”

  • The paper relies on this without support.

    Institutions can be held accountable for FERPA violations arising from faculty or students' unauthorized or unmonitored use of personal AI accounts.

    “because institutions can be held accountable for FERPA violations even if those violations arise from faculty or students' unauthorized or unmonitored use of personal accounts.”

  • The paper relies on this without support.

    The use of personal AI accounts by faculty and students for student data is widespread.

    “anecdotal evidence and survey data suggest that it is widespread, particularly among early adopters of generative AI tools.”

  • The paper relies on this without support.

    The rapid pace of technological change has outpaced the ability of independent researchers to conduct rigorous evaluations.

    “the rapid pace of technological change has outpaced the ability of independent researchers to conduct rigorous evaluations.”
6 of 6

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Search period extends beyond the paper's publication date

    Minor

    The paper reports searching publications through December 2026, but the issue is dated November-December 2025. A search completed for that issue could not include publications from 2026.

  • Vendor-only evidence receives two different rubric ratings

    Minor

    The evidence rubric rates consistent vendor-derived evidence as moderate, but also rates vendor-derived evidence only as weak. Consistent evidence derived only from vendors meets both descriptions, leaving its rating contradictory.

2 of 2

Cites

Flagged first, then by citations

Research works printed but not linked: 1

  1. Grajek, S., Pelletier, K. and Freeman, A. (2025) "AI procurement in higher education: Benefits and risks of emerging tools," EDUCAUSE Review [Preprint].

Authors
Akter, F., & Parvatha Reddy, K. K.
ORCID
Akter, Fatema
Venue
International Journal For Multidisciplinary Research
Publisher
International Journal for Multidisciplinary Research (IJFMR)
Published
2025-12-10
Volume
7
Issue
6
ISSN
2582-2160 (ISSN-L)
DOI
10.36948/ijfmr.2025.v07i06.84921
OpenAlex
W7172158214
Semantic Scholar
81c07ffde2eac37d3135be8a380a48bb20a3d226
Cited by
0
References
18
FWCI
0.00
Type
journal article
Open access
Licence
cc-by-sa
Locations
International Journal For Multidisciplinary Research (publishedVersion, cc-by-sa)
Topics
  • Physical Sciences › Computer Science › Information Systems › Artificial Intelligence in Education
  • Physical Sciences › Computer Science › Information Systems › Educational Innovations and Challenges
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
Keywords
higher education, technology selection, generative AI tools, UTAUT, technology acceptance, prompt engineering, adaptive learning, intelligent tutoring systems, FERPA compliance, accessibility standards, data residency, AI governance, academic integrity

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-19.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-19.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review