MétaCan
Menu
Back to cohort
Record W4415282304 · doi:10.3390/info16100904

Can We Trust AI Content Detection Tools for Critical Decision-Making?

2025· article· en· W4415282304 on OpenAlexaff
Tadesse G. Wakjira, Ibrahim A. Tijani, M. Shahria Alam, Mustafa Mashal, Khalad Hasan

Bibliographic record

VenueInformation · 2025
Typearticle
Languageen
FieldComputer Science
TopicExplainable Artificial Intelligence (XAI)
Canadian institutionsUniversity of British Columbia, Okanagan CampusGovernment of Manitoba
Fundersnot available
KeywordsGovernment (linguistics)Robustness (evolution)Precision and recallRecallReliability (semiconductor)Content analysis

Abstract

fetched live from OpenAlex

The rapid integration of artificial intelligence (AI) in content generation has encouraged the development of AI detection tools aimed at distinguishing between human- and AI-authored texts. These tools are increasingly adopted not only in academia but also in sensitive decision-making contexts, including candidate screening by hiring agencies in government and private sectors. This extensive reliance raises serious questions about their reliability, fairness, and appropriateness for high-stakes applications. This study evaluates the performance of six widely used AI content detection tools, namely Undetectable AI, Zerogpt.com, Zerogpt.net, Brandwell.ai, Gowinston.ai, and Crossplag, referred to as Tools A through F in this study. The assessment focused on the ability of the tools to identify human versus AI-generated content across multiple domains. Verified human-authored texts were gathered from reputable sources, including university websites, pre-ChatGPT publications in Nature and Science, government portals, and media outlets (e.g., BBC, US News). Complementary datasets of AI-generated texts were produced using ChatGPT-4o, encompassing coherent essays, nonsensical passages, and hybrid texts with grammatical errors, to test tool robustness. The results reveal significant performance limitations. The accuracy ranged from 14.3% (Tool B) to 71.4% (Tool D), with the precision and recall metrics showing inconsistent detection capabilities. The tools were also highly sensitive to minor textual modifications, where slight changes in phrasing could flip classifications between “AI-generated” and “human-authored.” Overall, the current AI detection tools lack the robustness and reliability needed for enforcing academic integrity or making employment-related decisions. The findings highlight an urgent need for more transparent, accurate, and context-aware frameworks before these tools can be responsibly incorporated into critical institutional or societal processes.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.000
metaresearch head score (Gemma)0.002
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Other design · Consensus signal: none
GenreCandidate signal: Methods · Consensus signal: none
Teacher disagreement score0.972
Threshold uncertainty score0.730

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0000.002
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.000
Science and technology studies0.0000.000
Scholarly communication0.0010.004
Open science0.0000.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.038
GPT teacher head0.327
Teacher spread0.290 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one teacher head, not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designOther design
Domainnot available
GenreMethods

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations3
Published2025
Admission routes1
Has abstractyes

Explore more

Same venueInformationSame topicExplainable Artificial Intelligence (XAI)French-language works237,207