Australia’s main research funder has barred peer reviewers from using artificial intelligence chatbots to produce feedback, following allegations that this had been happening.

The Australian Research Council published on the issue after applicants for grants of up to A$500,000 (£262,000) awarded under the Discovery Projects scheme reported spotting the “tell-tale” signs of ChatGPT in assessors’ comments.

The said the reports were a “generic regurgitation” of their applications with little evidence of critique, insight or assessment, and that one reviewer had even forgotten to remove the “regenerate response” prompt that appears at the bottom of all ChatGPT-created text.

The new guidance, published on 7 July, says peer reviewers “are required to preserve the principles of confidentiality”.

�鶹

Campus collection: AI transformers like ChatGPT are here, so what next?

“Release of material into generative AI tools constitutes a breach of confidentiality and peer reviewers…must not use generative AI as part of their assessment activities,” the policy says.

It adds that reviewers “are asked to provide detailed high quality, constructive assessments that assist the selection advisory committees to assess the merits of an application. The use of generative AI may compromise the integrity of the ARC’s peer review process by, for example, producing text that contains inappropriate content, such as generic comments and restatements of the application.”

�鶹

The policy says that, where the ARC suspects that reports are AI-generated, they will be removed from the review process, and that the ARC “may impose consequential actions in addition to any imposed by the employing institution”.

Australian researchers had suggested that the use of ChatGPT to write feedback was a symptom of the time pressure that academics in the country are under.

In terms of grant applicants, the ARC guidance says that, while AI “presents an opportunity to assist researchers in the crafting of grant proposals”, this “may raise issues around authorship and intellectual property including copyright. Content produced by generative AI may be based on the intellectual property of others or may also be factually incorrect.”

As such, the ARC “advises applicants to use caution in relation to the use of generative AI tools in developing their grant applications” and notes that universities are required to certify that all applicants “are responsible for the authorship and intellectual content of the application”.

�鶹

chris.havergal@timeshighereducation.com