A university framework for the responsible use of generative AI in research

May combine two claims.

Institutional responses to AI in research remain underdeveloped, unlike the abundant research on AI in undergraduate teaching.

“Unlike the abundant research on AI in undergraduate teaching (Crompton & Burke, 2023;Sullivan et al., 2023), institutional responses to AI in research remain underdeveloped (Popenici, 2023).”

• cites Popenici, 2023 sourceCould not verify

  The citation to Popenici attaches to the claim that institutional responses to AI in research remain underdeveloped, not to the comparison about undergraduate teaching. The 'AI and the aims of education' section discusses universities’ governance and changes to teaching and assessment, but not their responses to AI in research. Searches for research-specific institutional policies and responses found no passage addressing that claim. (LLM, conf=0.74)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim.

When AI is used in research with human participation or data, ethical implications require close consideration by Human Research Ethics Committees.

“When AI is used in research with human participation or data, ethical implications require close consideration by Human Research Ethics Committees, also known as Institutional Review Boards (Herington et al., 2023;Zhou & Chen, 2023).”

• cites Herington et al., 2023 sourcePartly supported

  The Abstract identifies ethical risks involving patients and says this paper focuses on clinicians’ and regulators’ obligations during AI deployment. It assigns researchers’ duties during data collection and development to a companion paper. The cited paper supports the general need to consider AI ethics, but the passages do not say that Human Research Ethics Committees or Institutional Review Boards should review research involving participants or their data. (LLM, conf=0.82)

  Experimental — accuracy not yet measured
• cites Zhou & Chen, 2023 sourceNot checked — no full text of the source

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

As Butson and Spronken-Smith (2024) have highlighted, it is essential that institutions consider contrasting and diverse perspectives related to the integration of AI in research tasks, because this is not merely a technological transition but also a significant ethical and methodological turning point.

“As Butson and Spronken-Smith (2024) have highlighted, it is also essential that institutions consider contrasting and diverse perspectives related to the integration of AI in research tasks, because 'this is not merely a technological transition but also a significant ethical and methodological turning point' (p.567).”

• cites Butson and Spronken-Smith (2024) sourceSupported

May combine two claims.

The legal aspects of copyright related to the outputs of AI in Australia have yet to be updated to address the use of AI and are inadequately regulated.

“The legal aspects of copyright related to the outputs of AI in Australia have yet to be updated to address the use of AI and are inadequately regulated (see Gaffar & Albarashdi, 2024;Samuelson, 2023).”

• cites Gaffar & Albarashdi, 2024 sourcePartly supported

  The section “The Dilemma of AI Authorship under Asian Law” says Australia lacks sufficient copyright protection for AI-created works because copyrightability requires human personality. The Conclusion likewise says Australian law protects only works created solely by humans. These passages support the claim of inadequate protection, but do not establish that Australian copyright law has yet to be updated to address AI. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites Samuelson, 2023 sourceNot checked — no full text of the source

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

Some scholars question the use of AI as infringing independent thought.

“Given that some scholars question the use of AI as infringing independent thought, there are additional questions needed to ensure the reputation of postgraduate work in this instance, as norms shift (Butson & Spronken-Smith, 2024).”

• cites Butson & Spronken-Smith, 2024 sourceSupported

Classified as probably not a substantial claim.

Misconceptions about what AI is and how it works remain common.

“As reported by Thompson et al. (2023), p. 3), 'there is still sensemaking work to be done, as misconceptions about what AI is and how it works remain common'.”

• cites Thompson et al. (2023), p. 3 sourcePartly supported

  In “Revisiting a research agenda for generative AI in higher education,” the paper says that misconceptions about generative AI remain common. The citing sentence omits “generative” from what it presents as a quotation, broadening the statement to AI generally. (LLM, conf=0.88)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S2 (checkable).

Inquiry and rigour to understand the challenges and benefits of AI updates will remain fundamental and ongoing issues central to academic endeavour and require rigorous scrutiny by institutions and researchers.

“For example, as AI developers and providers continue to address limitations and challenges related to model capabilities, inquiry and rigour to understand the challenges and benefits of AI updates to models and updates will remain fundamental and ongoing issues that are 'central to academic endeavour and require rigorous scrutiny' by institutions and researchers (Butson & Spronken-Smith, 2024, p.”

• cites Butson & Spronken-Smith, 2024 sourcePartly supported

  The Literature review (page not available in our copy) says AI’s limitations are central to academic endeavour and require rigorous scrutiny. The Concluding comments call for ongoing dialogue as AI evolves. Neither passage specifically discusses updates to AI models or scrutiny of the benefits and challenges of those updates, so the citation supports the general point but not that narrower application. (LLM, conf=0.85)

  Experimental — accuracy not yet measured

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

The two university stances include a rule that an AI tool must not be listed as an author, as this violates the principle of accountability for research outputs.

“An AI tool must not be listed as an author, as this violates the principle of accountability for research outputs.”

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

The Macquarie committee considered the risks and opportunities posed by AI to research integrity, processes and operations.

“They considered the risks and opportunities posed by AI to research integrity, processes and operations.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

Both the guidance note and the position statement proved strategically useful for the subsequent development of AI policy in 2025 and to align university thinking around consistent norms for the responsible use of AI in research.

“Both the guidance note and the position statement proved strategically useful for the subsequent development of AI policy in 2025 and to align university thinking around consistent norms for the responsible use of AI in research.”

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

The two university stances have minimum prescribed rules.

“They have minimum prescribed rules, such as:”

Classified as probably not a substantial claim.

The two university stances include a rule that explanations on how AI was used should be included in the Methods and/or Acknowledgements sections of a research output to be consistent with the policies of publishers and the principle of transparency.

“Explanations on how AI was used should be included in the Methods and/or Acknowledgements sections of a research output to be consistent with the policies of publishers and the principle of transparency.”

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

QUT began consultation in January 2023 by drafting a discussion paper titled 'Emerging AI tools and challenges for research integrity', which was circulated broadly to provoke discussion and feedback.

“QUT began consultation in January 2023 by drafting a discussion paper: Emerging AI tools and challenges for research integrity, which was then circulated broadly to provoke discussion and feedback.”

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

The two university stances support other existing responsible conduct of research policies.

“They support other existing responsible conduct of research policies.”

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

In 2023, when both universities were first writing their documents, no informed student voice was available for formal consultation.

“In 2023, when both universities were first writing their documents, no informed student voice was available for formal consultation.”

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

To create the Guidance Note, the Macquarie committee evaluated the current literature and engaged in cross-faculty discussions.

“To create the Guidance Note, the committee evaluated the current literature and engaged in crossfaculty discussions.”

May combine two claims.

Both Macquarie and QUT acknowledge tension between rigorous AI disclosure expectations and realistic supervisory practices, placing accountability largely on supervisors while recognising a need for clearer norms and ongoing education.

“Both Macquarie and QUT acknowledge this tension, placing accountability largely on supervisors while also recognising a need for clearer norms and ongoing education.”

Probably not a substantial claim: may not pass S3 (load-bearing). The quote only partly states this claim.

Macquarie produced a Guidance Note on the responsible use of generative AI in research.

“Macquarie produced a Guidance Note”

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

A Guidance Note outlining Macquarie University's position and recommendations regarding the use of AI in research was drafted.

“a Guidance Note outlining the University's position and recommendations regarding the use of AI in research was drafted.”

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

The QUT discussion paper outlined inherent risks AI could pose for responsible research practices, such as authorship, originality of publications and research outputs.

“It outlined the inherent risks AI could pose for responsible research practices, such as authorship, originality of publications and research outputs.”

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

The Guidance Note was submitted to the relevant committees for comment or endorsement.

“The Guidance Note was submitted to the relevant committees for comment or endorsement.”

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

QUT developed a Position Statement on the responsible use of generative AI in research.

“QUT developed a Position Statement”

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

Macquarie established a committee of stakeholders and consulted experts from each faculty and professional services.

“Macquarie established a committee of stakeholders and consulted experts from each faculty and professional services.”

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

For both QUT and Macquarie, consultation was coordinated by each university's Office of Research Ethics and Integrity (OREI).

“For both cases, consultation was coordinated by each university's Office of Research Ethics and Integrity (OREI).”

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

The two university stances encourage the use of AI where it promotes responsible research quality and impact.

“They encourage the use of AI where it promotes responsible research quality and impact.”

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

The two university stances are primarily principles-based so that readers can interpret their implementation within their own specific context.

“They are primarily principles-based so that the readers can interpret their implementation within their own specific context.”

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

During development at Macquarie, the research working group was instructed to consult with Higher Degree by Research (HDR) students and appropriate staff in their respective faculties.

“During development at Macquarie, in the three months of development of our Guidance Note, the research working group was instructed to consult with their Higher Degree by Research (HDR) students and appropriate staff in their respective faculties.”

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

Both QUT's Position Statement and Macquarie's Guidance Note articulate a university-level stance on the responsible use of generative AI in research.

“Both articulate a university-level stance on the responsible use of generative AI in research.”

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

Macquarie named its statement a 'Guidance Note', emphasising its role in highlighting extant policies and indicating how they apply to the responsible conduct of research.

“Macquarie named its statement a 'Guidance Note', emphasising its role in highlighting extant policies and indicating how they apply to the responsible conduct of research.”

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

Our university stances have been developed based on broad stakeholder feedback.

“Our university stances have been developed based on broad stakeholder feedback.”

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

QUT named its document a 'position statement' to emphasise that it provides definitive university-level advice.

“While QUT named its document a 'position statement' to emphasise that it provides definitive university-level advice”

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

Macquarie called its document a Guidance Note.

“It is worth noting that Macquarie called its document a Guidance Note for two reasons.”

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

The Macquarie committee was tasked with considering whether a new policy or targeted policy updates were required or whether changes to processes were deemed appropriate.

“The committee was tasked with considering whether a new policy or targeted policy updates were required or whether changes to processes were deemed appropriate.”

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

In 2023, the pool of researchers and practitioners with actual, thorough use of AI tools was quite narrow, leading to most consultation through faculty experts and those tasked with investigating these services.

“In 2023, this pool was quite narrow, leading to most consultation through faculty experts and those people tasked with investigating these services.”

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

The Macquarie Guidance Note was designed to be pedagogical, containing many footnotes with links to resources and explanations.

“Second, as a document which exists only in an advisory capacity, the format of the Macquarie Guidance Note was designed to be pedagogical, containing many footnotes with links to resources and explanations that would have been inappropriate in a more formal Procedure or Policy.”

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

QUT asked staff to reach out to well-informed students.

“QUT also asked staff to reach out to wellinformed students.”

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

The American Psychological Association requires authors to explicitly document AI use in their methods sections and to provide full AI outputs as supplemental materials.

“For instance, the American Psychological Association requires authors to explicitly document AI use in their methods sections and to provide full AI outputs as supplemental materials (2023).”

• cites American Psychological Association, 2023Not matched

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

Classified as probably not a substantial claim.

Researchers must document and transparently disclose AI integration into their research, detailing AI tools, methodologies, and potential biases.

“Researchers must document and transparently disclose AI integration into their research, detailing AI tools, methodologies, and potential biases (National Health and Medical Research Council, Australian Research Council, & Universities Australia, 2018, Principles 2, 3).”

• cites National Health and Medical Research Council, Australian Research Council, & Universities Australia, 2018, Principles 2, 3Not matched

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

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

Many journals currently require minimal disclosure following COPE guidelines.

“Many journals currently require minimal disclosure following COPE guidelines (Committee on Publication Ethics, 2024).”

• cites Committee on Publication Ethics, 2024Not matched

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

May combine two claims.

In Australia, the standard for doctoral degrees requires researchers' full responsibility and accountability for personal outputs, regardless of source, and sufficient cognitive skills for intellectual independence and critical thinking.

“In Australia, the standard for doctoral degrees requires researchers' full responsibility and accountability for personal outputs, regardless of source, and sufficient cognitive skills for intellectual independence and critical thinking (Australian Qualifications Framework Council, 2013).”

• cites Australian Qualifications Framework Council, 2013Not matched

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

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

The rapid evolution of AI complicates the enterprise software purchasing process, as products may change significantly before they are fully implemented.

“The rapid evolution of AI also complicates the enterprise software purchasing process, as products may change significantly before they are fully implemented.”

Probably not a substantial claim: may not pass S2 (checkable).

The required active supervision and written consent may be burdensome.

“The required 'active supervision and written consent' may be burdensome; thus, institutions must balance rigorous expectations with realistic supervisory practices.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

The continually changing technology makes maintaining a consistent and aligned university stance an ongoing task.

“The continually changing technology makes maintaining a consistent and aligned university stance an ongoing task.”

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

This acknowledgement-permissive mode ensures that content responsibility always remains with human authors.

“This acknowledgement-permissive mode ensures that content responsibility always remains with human authors.”

Classified as probably not a substantial claim.

In Australia, institutions have a responsibility to provide ongoing training and education that promotes and supports responsible research conduct for all researchers.

“In Australia, institutions have a responsibility to 'Provide ongoing training and education that promotes and supports responsible research conduct for all researchers' (National Health and Medical Research Council, Australian Research Council, & Universities Australia, 2018).”

• cites National Health and Medical Research Council, Australian Research Council, & Universities Australia, 2018Not matched

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

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.

Universities might deploy AI to reduce staff loads, an efficiency expected to extend to research integrity and ethics committees.

“AI may contribute to digital transformation efforts, similar to the push for paperless offices in the 1990s (Vrana & Singh, 2021), and universities might deploy AI to reduce staff loads, an efficiency expected to extend to research integrity and ethics committees.”

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

A 2024 Australian government review of higher education quality, accessibility and sustainability highlighted the potential of AI to boost research productivity and urged universities to lead in building new research infrastructure.

“A 2024 Australian government review of higher education quality, accessibility and sustainability highlighted the potential of AI to boost research productivity, urging universities to lead in building new research infrastructure (Department of Education, 2024).”

• cites Department of Education, 2024Not matched

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

Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing).

Macquarie University requires that all stages of a HDR or Honours student's research journey be well documented for data integrity and as evidence of originality and intellectual input if challenged.

“At Macquarie University, it is recommended that all stages of a students research journey be well documented both for data integrity purposes and to serve as evidence of originality and intellectual input if challenged.”

• cites Macquarie University Guidance Note, Section 2.3Not matched

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

May combine two claims.

Relevant ethical issues under the National Statement relate to bias and committee membership.

“Relevant ethical issues that fall under the scope of the National Statement relate to bias and committee membership.”

Classified as probably not a substantial claim.

AI can be used for subsidiary research publishing tasks, provided acknowledgement is included.

“They generally agree that while AI cannot be held responsible as authors, it can be used for subsidiary research publishing tasks (Nature Portfolio, 2023), provided acknowledgement is included.”

• cites Nature Portfolio, 2023Not matched

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

Probably not a substantial claim: may not pass S4 (based).

The framework is based on current literature and practical implementation plans.

“Our framework, based on current literature and practical implementation plans, serves as a foundational and adaptable model offering clarity and direction for research institutions.”

Classified as probably not a substantial claim.

A Guidance Note did not require full senate approval for updates, allowing it to be a responsive living document.

“First, pragmatically, a Guidance Note did not require full senate approval for updates, allowing for the note to be a responsive living document.”

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

Our examples illustrate that while the approach to developing a university position can vary according to each institution's needs and internal structure, coordinated consultation, discussion and feedback across university research communities is crucial to ensure all stakeholders are provided the opportunity to participate in the process.

“Our examples illustrate that while the approach to developing a university position can vary according to each institution's needs and internal structure, coordinated consultation, discussion and feedback across university research communities is crucial to ensure all stakeholders are provided the opportunity to participate in the process.”

May combine two claims.

Changing copyright standards for AI involvement may cause a published work to not meet the human author threshold, and thus may not be eligible for protection under the current law.

“Changing copyright standards for AI involvement may cause a published work to not meet the human author threshold, and thus may not be eligible for protection under the current law.”

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

Macquarie University requires that generative AI not be used without supervisory approval.

“Generative AI should not be used without supervisory approval”

• cites Macquarie University Guidance Note, Section 2.3Not matched

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

May combine two claims.

AI may not be inherently biased, but the data on which AI is trained and the interpretation of its results can lead to bias.

“While AI may not be inherently biased, the data upon which they are trained and the interpretation of their results can lead to bias.”

Classified as probably not a substantial claim.

The beginning of trust has to be an in-person contact.

“As Vinge (2006) predicted, '[T]he beginning of trust has to be an in-person contact'.(p.129)”

• cites Vinge (2006)Not matched

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

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

AI may contribute to digital transformation efforts in universities, similar to the push for paperless offices in the 1990s.

“AI may contribute to digital transformation efforts, similar to the push for paperless offices in the 1990s (Vrana & Singh, 2021), and universities might deploy AI to reduce staff loads, an efficiency expected to extend to research integrity and ethics committees.”

• cites Vrana & Singh, 2021 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Another issue in accord with the National Statement involves technical expertise on research ethics committees.

“Another issue in accord with the National Statement involves technical expertise on research ethics committees.”

• AssumptionRests on an assumption: Current research ethics committee membership may not include the expertise to ethically evaluate research with AI.

May combine two claims.

Postgraduate students have a dual identity: their work must be subject to assessment and degree standards, but as researchers they can be treated as academics and even equals.

“On the one hand, their work must be subject to assessment and degree standards, but as researchers, they can be treated as academics and even equals (Tanggaard & Wegener, 2017).”

• cites Tanggaard & Wegener, 2017 sourceNot checked — no full text of the source

May combine two claims.

The Macquarie Guidance note requires the active supervision and written consent of the supervisor, allowing the supervisor to govern the AI use of their student.

“The Macquarie Guidance note (Section 2.3) splits this Gordian knot by requiring the 'active supervision and written consent' of the supervisor, allowing the supervisor to actively and narrowly govern the AI use of their student.”

• cites Macquarie Guidance note (Section 2.3)Not matched

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

Classified as probably not a substantial claim.

AI cannot be held responsible as authors in academic publishing.

“They generally agree that while AI cannot be held responsible as authors, it can be used for subsidiary research publishing tasks (Nature Portfolio, 2023), provided acknowledgement is included.”

• cites Nature Portfolio, 2023Not matched

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

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

Human research ethics is guided by the National Statement on Ethical Conduct in Human Research, which outlines the ethical aspects of human research design, review and conduct.

“Human research ethics is guided by the National Statement on Ethical Conduct in Human Research (the National Statement), which outlines the ethical aspects of human research design, review and conduct (National Health and Medical Research Council, Australian Research Council, & Universities Australia, 2023).”

• cites National Health and Medical Research Council, Australian Research Council, & Universities Australia, 2023Not matched

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

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

This strategic framework then supports the development of a governance structure that guides institutional responses and adaptations over time.

“This strategic framework then supports the development of a governance structure that guides institutional responses and adaptations over time.”

Classified as probably not a substantial claim.

Under-reliance on AI for editing and polishing, resulting in extremely problematic first drafts, is a failure mode for postgraduate students.

“The second is the opposite problem: extremely problematic first drafts and under-reliance on AI for editing and polishing.”

Classified as probably not a substantial claim. May combine two claims.

AI literacy is fast becoming an essential part of research processes and a valuable industry skill.

“Moreover, AI literacy is fast becoming an essential part of research processes and a valuable industry skill (see Hutson, 2023;Prillaman, 2024;Van Noorden & Perkel, 2023).”

• cites Hutson, 2023;Prillaman, 2024;Van Noorden & Perkel, 2023Not matched

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

Probably not a substantial claim: may not pass S4 (based).

A written stance serves as a foundational guide, articulating the university's principles and position on the responsible use of AI in research.

“A written stance serves as a foundational guide, articulating the university's principles and position on the responsible use of AI in research.”

Classified as probably not a substantial claim.

Over-reliance on AI without the student engaging in critical thinking is a failure mode for postgraduate students.

“The first is over-reliance on AI without the student engaging in any critical thinking, where 'the AI did my thesis for me' is the worst outcome.”

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

The Australian Research Council acknowledges that while AI presents opportunities for research and grant proposal writing, the responsibility for errors falls on researchers and their institutions.

“The Australian Research Council acknowledges that while AI presents opportunities for research and grant proposal writing, the responsibility for errors falls on researchers and their institutions.”

• cites Australian Research Council Research Policy Branch, 2023Not matched

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

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

Mitigating risks is a primary priority, exemplified by the European Union's strict legislation on transparency, bias, privacy and copyright principles.

“Mitigating risks is a primary priority, exemplified by the European Union's strict legislation on transparency, bias, privacy and copyright principles (Gibney, 2024).”

• cites Gibney, 2024 sourceNot checked — no full text of the source

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

Macquarie University requires supervisors to closely monitor the use of generative AI by their research students.

“Supervisors must closely monitor the use of Generative AI by their research students.”

• cites Macquarie University Guidance Note, Section 2.3Not matched

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

Classified as probably not a substantial claim.

Combined broad and specific consultation ensures that institutions not only address the challenges but also explore the opportunities AI provides.

“Combined broad and specific consultation ensures that institutions not only address the challenges but also explore the opportunities AI provides.”

Classified as probably not a substantial claim.

A written stance is not a policy itself, but rather informs the development of specific institutional policies and procedures that establish enforceable guidelines.

“However, it is not a policy itself, but rather informs the development of specific institutional policies and procedures that establish enforceable guidelines.”

Probably not a substantial claim: may not pass S4 (based).

The uncertainty of how AI will ultimately impact the research sector is driven by complex research regulatory environments, rapid innovation, and differing norms between research disciplines, universities and countries.

“This is a significant challenge due to the uncertainty of how AI will ultimately impact the research sector, which is driven by complex research regulatory environments, rapid innovation, and differing norms between research disciplines, universities and countries.”

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

The level of active attention given by the supervisor is up to the supervisor's discretion.

“The level of active attention given by the supervisor is up to the supervisor's discretion.”

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

Mollick (2025) described using AI to detect statistical errors in journal articles.

“For example, after Mollick (2025) described using AI to detect statistical errors in journal articles, Macquarie updated its Peer Review policy to permit reviewers to use AI tools that support their judgement, provided this does not violate confidentiality or commissioning guidelines.”

• cites Mollick (2025)Not matched

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

Classified as probably not a substantial claim.

These cases demonstrate how institutions can adapt policies to reflect emerging scholarly practices through regular review.

“These cases demonstrate how institutions can adapt policies to reflect emerging scholarly practices through regular review.”

• AssumptionRests on an assumption: Macquarie updated its Peer Review policy to permit reviewers to use AI tools that support their judgement, provided this does not violate confidentiality or commissioning guidelines.
• AssumptionRests on an assumption: QUT is implementing additional training based on stakeholder feedback.
• AssumptionRests on an assumption: Macquarie is tracking these conversations through checking compliance at Data Management Plan submission stage, along with ad hoc conversations with supervisors during graduate training.
• AssumptionRests on an assumption: QUT is tracking the effectiveness of its training through training evaluation surveys.

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

At QUT, any use of AI, including text editing for readability and grammar, must be disclosed in the thesis acknowledgement and a detailed written description documenting the extent of AI use and how it has been used is required.

“Any use of AI, including text editing for readability and grammar, must be disclosed in the thesis acknowledgement and a detailed written description documenting the extent of AI use and how it has been used is also required (QUT Digital Workplace”

• cites QUT Digital Workplace "AI in Research: responsible use", personal communication, 2025Not matched

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

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

At QUT, the position statement focuses on the university's commitment to research integrity.

“At QUT, the position statement focuses on the university's commitment to research integrity.”

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

The choice of terms of 'active supervision' is to emphasise the supervisor's responsibility of this component of research training.

“The choice of terms of 'active supervision' is to emphasise the supervisor's responsibility of this component of research training.”

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

Other practitioners have produced frameworks for universities dealing with AI broadly, including for undergraduate policy.

“While other practitioners have produced frameworks for universities dealing with AI broadly (for undergraduate policy, see Luo, 2024), this paper specifically addresses the research policy context.”

• cites Luo, 2024Not matched

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

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

The Macquarie Guidance Note explicitly empowers supervisors to make the guidelines specific per student because students and supervisors have different needs and expectations when it comes to writing and AI use.

“The Macquarie Guidance Note chose to explicitly empower supervisors to make the guidelines specific per student as students and supervisors have different needs and expectations when it comes to writing and AI use.”

• cites Macquarie Guidance NoteNot matched

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

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

The Macquarie Guidance Note requires supervisors to have iterative exposure to their students' work across their entire research degree and to closely monitor research students' use of Generative AI.

“Supervisors should have iterative exposure to their students work across their entire research degree and ensure their research progression (including critical judgement or originality) is observed and well documented.Supervisors must closely monitor the use of Generative AI by their research students”

• cites Macquarie Guidance Note, section 2.3Not matched

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

Classified as probably not a substantial claim. May combine two claims.

The evaluation process provides a review mechanism to inform stakeholders of policy effectiveness, ensures the currency of the university's stance and informs new directions and updates to the framework.

“The evaluation process, therefore, provides a review mechanism to inform stakeholders of policy effectiveness, ensures the currency of the university's stance and informs new directions and updates to the framework.”

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

The phrasing in the Macquarie Guidance Note is an acknowledgement that as researchers in training, the consequences of AI use will fall on the supervisor.

“This phrasing is an acknowledgement that as researchers in training, the consequences of AI use will fall on the supervisor.”