What happened
The West Australian reports that Deloitte’s University Sector Review: Rapid Cost Benefit Analysis, prepared for the Western Australian Government’s university-merger push, contains multiple citation errors despite Deloitte stating that professionals comprehensively checked its AI-assisted work. Reported problems include incorrect research titles, publication dates, author credits and a mismatched link. Deloitte acknowledged that it was reviewing the but declined public comment.
The West Australian reports that Deloitte disclosed using AI to support research collation, information synthesis and editorial proofing in its University Sector Review: Rapid Cost Benefit Analysis. Deloitte’s document also stated that all AI-assisted outputs underwent a comprehensive quality-assurance review by Deloitte professionals for accuracy, relevance and appropriateness.
According to The West Australian, the report’s references include a 2022 article attributed to L. T. Tran whose link leads to a different article, a 2023 consultants’ report cited under the wrong title, and a 2018 Grattan Institute report with one author replaced by an institute staffer. The outlet also reports numerous incorrect publication dates.
The errors concern supporting findings about a possible merger involving the University of Western Australia, Curtin University and Murdoch University. Deloitte was alerted to some errors by social media and The West Australian, and acknowledged that it was reviewing the citations. It declined public comment, according to the outlet.
The supplied report does not independently establish that caused every identified error. The West Australian quotes an anonymous tertiary educator who attributed the pattern to large language models, but that assessment is not independently verified here. The source also does not identify the model or models used, provide a complete error count, or show that the report’s underlying calculations are incorrect.
Source details: thewest.com.au ↗
Why it matters
The report is being used in a consequential public debate about restructuring Western Australia’s universities. Citation errors do not by themselves establish that Deloitte’s merger analysis is wrong, but they weaken confidence in the evidence trail supporting its conclusions. The episode also shows why disclosure of AI use is not a substitute for independently checking sources, especially when consultants’ reports may influence government decisions, institutions and staff.
The report sits within a government-backed process that could affect university governance, finances, staff and students. Its references are therefore more than editorial details: they are part of the evidentiary basis readers may use to assess the proposed merger.
A citation that points to the wrong paper or gives incorrect authorship can make verification difficult even if the broader conclusion happens to remain valid. The supplied reporting does not show that the errors changed Deloitte’s cost-benefit calculations, so the practical impact on the merger case remains unknown.
The episode is particularly relevant to professional use of AI because Deloitte said human quality assurance had occurred. The reported discrepancies suggest that a review process may need to check source identity, bibliographic details and whether each citation actually supports the associated claim, rather than relying on a general accuracy review.
The West Australian also reports that Deloitte previously apologized and repaid nearly $100,000 after an AI-assisted federal report contained similar citation problems and references to nonexistent documents. That history increases the significance of the new reporting, although the supplied material does not independently compare the two reports’ error rates or methods.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
crm_get_transaction(id='4092').Why can ethical evaluation not be reduced to one model score?
What to watch next
The key unresolved question is whether Deloitte’s review of the leads to corrections or changes to the report’s analysis. It is also not independently confirmed from the supplied material whether any underlying data, calculations or merger conclusions are materially affected. The source does not identify which AI system was used, when the report was delivered, or whether the Western Australian Government will commission an independent review.
Watch for Deloitte’s citation review, corrections, or a reissued version of the university report. The source does not state whether a correction timetable has been set.
Watch for an independent assessment of the report’s calculations, source data and merger conclusions. Whether the errors are confined to references or extend into substantive analysis is not established by the supplied reporting.
Watch for the Western Australian Government and the affected universities to explain whether they will continue relying on the report, seek additional evidence, or commission separate analysis.
The AI system used, the prompts or workflow involved, the full number of citation errors, the report’s delivery date and any documented effect on recommendations are all unknown from the supplied material.