{"id":"b0b9310c-0871-49c1-859a-6742d81a08f5","arxiv_id":"2411.13946","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A systematic review of 45 papers finds augmented reality acceptance research in training and education is dominated by TAM-based models with many idiosyncratic extensions and no validated general model.","lead":"This paper reviews 45 studies on why people accept augmented reality in training and education, and finds that most use the Technology Acceptance Model with many one-off extra variables. It is a map of a young research field, useful mainly to researchers planning new acceptance studies.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Step 2's TAM/UTAUT-only filter makes the headline conclusion 'TAM is most frequently used' partly an artifact of inclusion criteria.","rationale":"The strongest claim is essentially about comprehensiveness and TAM dominance. The most load-bearing condition for that claim is that the search identifies all relevant work. The protocol in Section III-B has two restrictions that each exclude relevant studies: title-only search (step 1) and TAM/UTAUT-only model filter (step 2). The second is especially damaging because it directly creates the TAM dominance finding: if a paper uses 'adoption' instead of 'acceptance' or a non-TAM theory, it is invisible. The reader's weakest assumption identifies the same risk, and I agree. A concrete replication with broader keywords would settle whether the 45-paper corpus is materially incomplete. If the corpus grows or the model mix changes, the conclusion about no generalizable model may still hold, but the specific claims of being the first comprehensive inventory and of TAM dominance would need qualification. This is not a fatal flaw; the review could be corrected, so the conditional verdict stands. Credit: the theoretical background is standard, the sample tables provide a useful starting point, and the paper honestly notes that many variables are one-off extensions. However, the absence of a PRISMA-style flow diagram, date limits, and quality appraisal increases the risk that the reported numbers are not robust.","tokens_in":12534,"tokens_out":4191,"duration_ms":35303,"concrete_test":"Replicate the search on the same seven databases using TITLE-ABS-KEY('augmented reality') AND TITLE-ABS-KEY('acceptance' OR 'adoption' OR 'intention to use' OR 'user experience' OR 'technology acceptance model') AND TITLE-ABS-KEY('education' OR 'training' OR 'learning'), with no model filter at step 2. Count the resulting corpus and the share of TAM-based versus non-TAM-based papers; if new papers appear or the TAM share falls materially below the reported near-totality, the headline claim is an artifact of the filter.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that this is the first comprehensive inventory and that TAM dominates—depends on the corpus being representative of all AR acceptance research in training and education. Section III-B step 2 filters the 204 title-matched records down to 45 by requiring that papers 'refer to a specific technology acceptance model, i.e., the TAM or UTAUT and its variants.' This inclusion criterion guarantees that TAM/UTAUT-family models are overrepresented: any study framed around adoption, user experience, or intention to use without naming TAM/UTAUT is excluded, as are studies grounded in other acceptance theories (e.g., IDT, SCT, TPB). The subsequent frequency counts (18 TAM-based, 2 UTAUT-based among the 22 training/education papers) are therefore not independent evidence of TAM dominance; they are a consequence of the screening rule. The title-only restriction in step 1 compounds this by missing papers whose titles use alternative terminology. Additionally, the paper's internal counts do not reconcile (for example, Section IV-E reports 16 training/education variables, then lists 11 undefined plus 10 defined), suggesting extraction errors that further weaken the inventory's reliability.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports a systematic literature review of technology acceptance research on augmented reality (AR) in training and education. The authors searched seven bibliographic databases, identified 204 articles with 'acceptance' and 'augmented reality' in the title, filtered to 45 articles that reference a named acceptance model (TAM or UTAUT family), and further narrowed to 22 articles specifically on training and education. They extract the research model, methods, sample, and extension variables from each study, and report that TAM and its core constructs dominate the field, that most models are one-off extensions with no generalizable validated model, and that there is a research gap in corporate training contexts. The central contribution claimed is a comprehensive inventory of AR acceptance models and variables for this application field.","tokens_in":12729,"tokens_out":3494,"duration_ms":33517,"significance":"If the inventory is valid, the paper would provide a useful first map of a young and fragmented research area: it would document the predominance of TAM-based models, list the specific variables proposed in AR acceptance extensions, and identify the absence of a validated generalizable model. The paper's explicit acknowledgment that most extension variables are used only once and that the field lacks validated models is a fair and valuable synthesis. However, the significance currently hinges on the representativeness of the 45- and 22-paper corpora, and the reported counts contain internal inconsistencies that prevent the inventory from being used as a reliable reference. The paper does not provide machine-checked data, a reproducible extraction protocol, or inter-coder validation, so the evidentiary basis for its headline claims is not fully auditable.","major_comments":[{"comment":"The inclusion criterion in Step 2 (Section III-B) requires that papers 'refer to a specific technology acceptance model, i.e., the TAM or UTAUT and its variants.' This makes the later finding that 'the TAM and its core variables are the most frequently used models' (Section V) partly an artifact of the screening rule: any study framed around adoption, user experience, or intention without naming TAM/UTAUT, or grounded in other theories such as IDT, TPB, or SCT, is systematically excluded. To support the headline claim of TAM dominance, the authors must either broaden the inclusion criteria to all acceptance theories and relevant terminology, or explicitly justify and report the effect of the TAM/UTAUT-only filter, for example by comparing retrieval counts under alternative search strategies.","section":"III-B, Step 2"},{"comment":"The systematic review protocol lacks a date range, a documented inclusion/exclusion protocol beyond the title and model filters, a quality appraisal step, and inter-coder validation. Without these, the claim in Section III-A that 'there is no work to date that more comprehensively examines' the field cannot be substantiated, since the search window is unbounded and the screening decisions are not auditable. The authors should specify the search date, the exact query per database, the exclusion reasons at each stage (ideally with a PRISMA-style flow diagram), and a procedure for resolving coding disagreements.","section":"III-B overall"},{"comment":"The variable counts in Section IV-E do not reconcile. The text states that '16 AR acceptance variables have been identified' in training and education, but then lists five undefined variables (Teaching Experience, Technology Experiences, Characteristics of the system, Information Experience, Information Literacy), eight moderators from [35] (Duration of Use, Perceived Exertion, Emotion, Attachment, Harm, Perceived Change, Movement, Anxiety) while calling them 'six variables', and ten defined variables in Table II, for a total of at least 23. The abstract's total of 34 acceptance variables is also inconsistent with the sum of the section counts. These discrepancies undermine the reliability of the inventory and must be corrected with a single reconciled table of all variables and their assignment to papers.","section":"IV-E"},{"comment":"The model counts in Section IV-D are internally inconsistent. The text reports 18 TAM-based articles and 2 UTAUT-based articles among the 22 training/education papers, and then separately mentions 'four papers did not propose an extension' and 'another last article did not refer to any existing research model.' These numbers do not sum to 22, and it is unclear whether the four non-extension papers are a subset of the 18 TAM papers. The authors should provide a per-paper classification table (model used, whether extended, and which variables were added) so the frequency claims can be verified.","section":"IV-D"},{"comment":"The novelty claim that no prior work 'more comprehensively examines' AR acceptance models in training and education is asserted rather than demonstrated. No existing AR-specific or education-specific technology acceptance reviews are cited or compared, and because no date range is imposed the claim cannot be checked. The authors should either cite and discuss the closest prior reviews and explain the incremental contribution, or soften the claim to a statement about the scope of their own search.","section":"III-A"}],"minor_comments":[{"comment":"The text contains numerous typos and residual editing errors, including 'In our t he literature review' (Section III-A), 'reseach' (Section IV-D), 'Exensions' (Section IV-E heading), 'ad well' (Section V), 'Confernce' (reference [28]), and 'Volutariness' (Section II-B2). A careful proofreading pass is needed.","section":"Throughout"},{"comment":"The tables list variables with frequency counts, but the mapping between the frequencies and the cited sources is not always transparent; for example, 'Technology Optimism (2)' and 'Technology Innovativeness (2)' cite two sources each, but the text does not explain how the frequency was counted when a paper uses multiple variables. Adding a column for the specific papers that used each variable would improve auditability.","section":"IV-E, Tables I and II"},{"comment":"The statement that 'No article investigated the use of AR glasses' contradicts the inclusion of reference [28] (an acceptance model for smart-glasses-based tourism AR) if that paper is among the 45; the authors should clarify whether this statement applies only to the 22 training/education papers.","section":"IV-A"},{"comment":"The sentence 'These six variables have not been further defined' lists eight variables; please correct the count or the list.","section":"IV-E, reference [35]"},{"comment":"The appendix lists 22 papers, but 'J.-H. Loand Y.-F. Lai (2018)' contains a spacing typo, and some entries have inconsistent year formatting (e.g., [46] is listed as 2020 in the text but 2022 in the reference list); please standardize the year and author formatting.","section":"Appendix, Table III"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses a relevant and timely topic for a human-computer interaction / educational technology audience, and the descriptive synthesis of AR acceptance models could be a useful community resource. However, the methodological filters in the selection process and the unresolved count inconsistencies mean the paper is not yet ready for publication. The authors should be asked to re-run or at least re-report their search with a transparent protocol, reconcile all numbers, and temper the novelty claim. The topic fits the journal's scope, but the review must meet the standards of a systematic literature review."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this is a useful but small-bore mapping of AR acceptance research in training and education, and its central conclusion has a built-in bias. It's not a waste of time, but it needs methodological cleanup before the numbers can be trusted.\n\nWhat's genuinely new: it's the only review I know that specifically tallies acceptance models and extension variables in AR training/education. The observation that nearly every study invents its own one-off extensions and that no generalizable model has emerged is a real takeaway. The paper is clearly written and transparent about its three-step selection process, and the appendix listing the 22 training/education papers is handy for anyone starting in this area.\n\nNow the soft spots, and they're in proportion. The screening rule in Section III-B step 2 only keeps papers that explicitly refer to TAM or UTAUT and variants. That filter guarantees that TAM/UTAUT-family models dominate the corpus. So the conclusion \"TAM is the most frequently used model\" is not independent evidence; it's largely a restatement of the inclusion criterion. That's a load-bearing flaw, not a minor one. The title-only search in step 1 is also narrow and may miss papers using terms like \"adoption\" or \"intention to use.\" There's no date range, no quality appraisal, no inter-coder validation, and no extraction table showing what was pulled from each paper.\n\nThe internal mathematics are a bigger problem than the reader's report suggested. Section IV-D reports 18 TAM-based and 2 UTAUT-based papers, then mentions four original-model papers and one with no model; that doesn't sum to the 22 training/education papers. Section IV-E says 19 training/education papers had extensions, but also says 12 proposed extended models, and later mentions 16 variables while listing 5 undefined plus 8 moderators plus 10 defined in Table II, which is 23. The abstract says 33 models and 34 variables over 45 papers, but the body's numbers never reconcile. These are not cosmetic. They mean the inventory cannot be audited.\n\nWho is this for? Someone planning an AR acceptance study in education who wants a starting bibliography of primary studies. It's a rough map, not a trustworthy quantitative summary. With a proper protocol, cleaned counts, and a discussion of how the filter shapes the headline, it could become a decent niche review.\n\nI'd send it to peer review, but only after substantial revision. The topic is legitimate and the field needs this kind of synthesis, but not with the method as it stands.","headline":"A modest, honest inventory of a narrow subfield, but its headline claim that TAM dominates is partly built into its own screening rule, and the reported counts don't reconcile.","tokens_in":13224,"tokens_out":3135,"would_cite":false,"duration_ms":31722,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The first systematic inventory of augmented-reality acceptance research in training and education finds a young, TAM-dominated field with no validated generalizable model.","keywords":["technology acceptance","augmented reality","TAM","UTAUT","systematic literature review","training and education","perceived usefulness","perceived ease of use"],"falsifier":"Run the same seven databases with a search that does not require the terms to be in the title, and that includes non-TAM theories such as the Diffusion of Innovations or Social Cognitive Theory; if this produces substantially more than 45 relevant AR acceptance papers, or many training/education papers outside the TAM family, the claimed comprehensiveness and TAM dominance would be overstated.","tokens_in":12341,"feed_emoji":"🎓","tokens_out":6487,"duration_ms":56181,"temperature":0.7,"pith_summary":"This paper asks what acceptance models and variables researchers use when studying augmented reality (AR) in training and education, and answers with a systematic review of the literature. The authors searched seven databases for papers with 'acceptance' and 'augmented reality' in the title, then kept only those using a technology acceptance model, ending with 45 papers, 22 of them in training and education. They report that the Technology Acceptance Model (TAM) and its core variables dominate the field, that three-quarters of the papers extend models with their own variables, and that no generally accepted, validated AR acceptance model has yet emerged. The review matters because it gives researchers a map of the current fragmentary state of the field and a starting point for validating existing models like MARAM across contexts.","feed_headline":"TAM dominates AR acceptance research in training and education","feed_subtitle":"A systematic review of 45 papers finds a young, TAM-based field with no validated generalizable model.","key_machinery":"The machinery is a three-step systematic literature selection protocol: first, search seven databases for titles containing both 'acceptance' and 'augmented reality,' yielding 204 papers; second, filter to papers that explicitly adopt a TAM/UTAUT-family model, yielding 45; third, filter to AR applications in training and education, yielding 22. The authors then code each paper on research objectives, sample data, research methods, the underlying acceptance model, and model extensions; the coding breakdown is what produces the headline frequency counts and the conclusion that TAM and its core variables are the most frequent theoretical foundation.","core_discovery":"The central claim is that technology acceptance research on augmented reality in training and education is still a young and TAM-dominated field. Across all application areas the authors identify 45 papers and 33 different acceptance models; within training and education, 18 of 22 papers build on TAM and its variants, while only two use UTAUT. The review finds 34 extension variables, but most are used by a single paper, and the authors conclude that there is no generalizable AR acceptance model that has been sufficiently validated in different application areas. The paper establishes a systematic inventory of the models, variables, samples, and methods in this niche, and positions it as a grounding for further analysis.","pith_inferences":["An implication the authors leave implicit: the low reuse of extension variables suggests many proposed 'AR acceptance' factors may be context-specific artifacts, and a meta-analysis across the 22 training/education studies could test which variables actually replicate as significant predictors.","The title-based search strategy could be checked directly by repeating the same seven-database search with broader terms like 'adoption,' 'user experience,' and 'intention to use,' and without the TAM/UTAUT filter; this would tell us whether the 45-paper corpus undercounts the field, though we expect the TAM-dominance ordering to persist.","The finding that no paper investigated AR glasses points to a device-form-factor moderator that the current literature cannot assess; a future review or empirical study comparing acceptance of mobile AR, head-mounted displays, and smart glasses would extend the inventory in a natural way."],"forward_implications":["Researchers designing AR training studies can treat TAM's Perceived Usefulness and Perceived Ease of Use as the default core constructs, since they are the most consistently integrated components across all 45 papers.","The absence of a validated generalizable model means the field has not yet settled on a standard model; the next step implied by the paper is to validate specialized models such as MARAM in new contexts rather than propose yet another extension.","Corporate and industrial AR training is a visible gap: most of the 22 training/education papers focus on academic teaching, so empirical acceptance studies outside schools and universities would address a documented lack.","Because three-quarters of the 45 papers extend an acceptance model with their own variables and most variables appear only once, the literature shows a pattern of model proliferation rather than cumulative validation."],"supporting_citations":[{"why":"Supplies the definition of augmented reality that frames the review's object of study.","marker":"[3]"},{"why":"The original Technology Acceptance Model, the dominant model counted across the 45 papers.","marker":"[7]"},{"why":"The UTAUT model, the second-most-used theoretical foundation identified in the corpus.","marker":"[8]"},{"why":"Defines the TAM variables and relationships, the target of most extension attempts in the review.","marker":"[10]"},{"why":"Provides the canonical definitions of Perceived Usefulness and Perceived Ease of Use that the frequency counts rely on.","marker":"[11]"},{"why":"The Mobile AR Acceptance Model, the exemplar specialized model the authors recommend for broader validation.","marker":"[42]"},{"why":"An example of a model extension mixing TAM, UTAUT, and IS Success Model, illustrating the proliferation pattern.","marker":"[23]"},{"why":"A prior review of technology acceptance models that the paper positions its own gap claim against.","marker":"[6]"}],"fun_headline_variants":["TAM dominates AR acceptance studies in training and education","AR training acceptance research: only 22 papers, TAM leads","No validated AR acceptance model for training and education yet","Systematic review of 45 AR acceptance papers in education","18 of 22 AR training studies use TAM, review says"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole inventory rests on the assumption that relevant studies keep both 'acceptance' and 'augmented reality' in their titles and that the only acceptable theoretical basis is a TAM/UTAUT-family model, so studies using different words or different acceptance theories are invisible to the search.","fun_headline_variants_meta":{"raw":{"variants":["TAM dominates AR acceptance studies in training and education","AR training acceptance research: only 22 papers, TAM leads","No validated AR acceptance model for training and education yet","Systematic review of 45 AR acceptance papers in education","18 of 22 AR training studies use TAM, review says"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000657,"raw_usage":{"total_tokens":2935,"prompt_tokens":801,"completion_tokens":2134,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":417,"completion_tokens_details":{"reasoning_tokens":2065}},"tokens_in":417,"tokens_out":2134,"duration_ms":15453,"temperature":1.0,"reasoning_tokens":2065,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:41:58.571573+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same seven databases with a search that does not require the terms to be in the title, and that includes non-TAM theories such as the Diffusion of Innovations or Social Cognitive Theory; if this produces substantially more than 45 relevant AR acceptance papers, or many training/education papers outside the TAM family, the claimed comprehensiveness and TAM dominance would be overstated.","supporting_citations":[{"cited_title":"A Survey of Augmented Reality","cited_arxiv_id":null,"evidence_quote":"Supplies the definition of augmented reality that frames the review's object of study."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The original Technology Acceptance Model, the dominant model counted across the 45 papers."},{"cited_title":"User Acceptance of Information Technology: Toward a Uni- fied View","cited_arxiv_id":null,"evidence_quote":"The UTAUT model, the second-most-used theoretical foundation identified in the corpus."},{"cited_title":"User Ac- ceptance of Computer Technology: A Comparison of Two Theoretical Models","cited_arxiv_id":null,"evidence_quote":"Defines the TAM variables and relationships, the target of most extension attempts in the review."},{"cited_title":"Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology","cited_arxiv_id":null,"evidence_quote":"Provides the canonical definitions of Perceived Usefulness and Perceived Ease of Use that the frequency counts rely on."},{"cited_title":"Koutromanos and T","cited_arxiv_id":null,"evidence_quote":"The Mobile AR Acceptance Model, the exemplar specialized model the authors recommend for broader validation."},{"cited_title":"A Theoretical Model of Augmented Reality Acceptance","cited_arxiv_id":null,"evidence_quote":"An example of a model extension mixing TAM, UTAUT, and IS Success Model, illustrating the proliferation pattern."},{"cited_title":"A review of technology acceptance and adoption models and theories","cited_arxiv_id":null,"evidence_quote":"A prior review of technology acceptance models that the paper positions its own gap claim against."}],"review_version":1}