Use of AI in Research
AI is constantly changing, so we encourage staff to regularly check in for updates and changes on the use of AI (incl. Generative AI) at UTS.
AI can mean different things to different people (refer to the UTS GenAI Guidelines for Users for more details). Generally, AI is good at recognising patterns from learned knowledge and analysing, summarising, or reorganising information based on set criteria. This makes it valuable for researchers in summarising large bodies of research, creating predictions and analyses from datasets, transcribing and translating information, and even checking grammar and tone in writing.
CoPilot (when signed into a UTS account) is currently the only UTS approved AI tool for any internal, sensitive, or confidential data.
Unless published, all research data should be considered as sensitive data, and in some instances confidential data.
Before jumping straight into using AI tools, there are a few things that researchers need to consider:
- All research is still beholden to the Research Policy, the use of AI also must not breach UTS Policies (see the Policies, Procedures and Guidelines for AI in Research page for more information and guidance) or facilitate any illegal activity.
- Have you considered whether your intended AI tool is right for your research? To learn more about the potential risks, ethical considerations, and integrity considerations associated with AI you can check out the Selecting AI Tools for Research page. If you want assistance, you can ask a Research Integrity Advisor.
- Remember that although AI can do many things, a tool is only as good as the expert who uses it. So, if you use AI to help with your tasks, as the expert, it is your responsibility to check the quality and integrity of the work produced using AI.
What can I do in CoPilot?
If you are using internal, sensitive or confidential data here’s some of the things you can do using CoPilot.
- Summarising papers and documents: CoPilot can summarise documents in Word, or across multiple files in OneDrive. For research analysis, apply disciplinary norms for validation and identify relevant emerging literature.
- Idea generation and drafting content: CoPilot can provide suggestions and draft content in Word, in Outlook, and in PowerPoint
- Data analysis: CoPilot can assist with data analysis, suggesting formulas, identifying insights, cleaning data or help with conditional formatting in Excel. It can also analyse large tabular datasets.
- Transcription / translation of interviews and documents: Microsoft Word, Outlook, PowerPoint and OneNote have a speech-to-text function built in, Microsoft Stream can transcribe pre-existing videos or recorded Teams meetings, and CoPilot can act as a live interpreter in Teams.
- Organising notes and meetings: CoPilot can summarise information from your Outlook Calendar, from meeting-related documents and emails in preparation for upcoming meetings. It can also summarise real-time meetings, create meeting minutes and task lists, and act as a meeting facilitator and moderate discussions.
- Research administration: CoPilot can help automate research tasks (alongside Power Automate), and can analyse and create Sharepoint Lists and Microsoft Forms.
- Finding documents and information: CoPilot can search through all documents within the UTS Microsoft ecosystem that you have access to – this means you can ask it questions about your Sharepoint pages, Teams sites, Documents on OneDrive and Outlook emails.
How do I know if the research data I am using is public or not?
If you are unsure of the data security level for your work, you can refer to the Library’s Research Data Management Page (they provide a 4-step process to classify, store and archive your research data), the UTS Information Security Classification Standard, or the UTS Research Data Classifier Decision Support Tool.
If you have concerns about any breaches of internal, sensitive or confidential research data you can check the Data Breach Policy, talk to a Research Integrity Advisor or email data.breach@uts.edu.au.
Public Data includes published research data and publications. It is information that is openly available to the general public.
- Generation of high-level ideas, or basic searches to learn about new research areas.
- Searching through publicly available data, or through tools embedded in existing research databases.
- Translation of everyday conversation with others (with appropriate consent), or public documents.
Internal Data includes soon to be published research content including but not limited to research data, project plans, research results, and draft manuscripts pending publication.
- Manuscripts pending publication may be considered as internal data, as they are about to be released publicly. However, if in doubt it is better to treat them as sensitive.
Sensitive Data includes research data including personal information from low and nil/negligible risk research, or from research not requiring approval from the HREC, faculty or unit performance reports, sensitive committee minutes, exam material or results.
- If not confidential data, research data is considered sensitive until ready to be published, including for research not requiring ethics approvals.
- Research methodologies or detailed research project / program plans. This includes systematic reviews, data analysis methods and results, and creation/collection of research data.
- Transcription or translation of participant interviews or meetings to be used for research analysis, or containing sensitive information (including project meetings, research discussions, interviewee personal data etc).
Confidential Data includes research data from high-risk research, requiring approval from HREC, health/medical data, university (including faculty/unit) forward business plans, commercial in-confidence information, or data owned by a third party.
- Any data owned by a third party (unless it is public data) is likely to be considered as confidential.
- Any research data from projects requiring ethics approval.
- Any research data from projects in high-risk or critical technologies areas.
What safety precautions is UTS taking?
UTS aims to ensure staff safety from AI tool risks. Currently, CoPilot is the only tool fully vetted by ITU for data security and safety, ensuring all data remains within UTS systems.
Associated risks of using AI tools:
- Security risks: AI tools may harm UTS systems and data, leading to corruption or loss of sensitive data and system issues.
- Legal risks: AI tools may cause privacy breaches, intellectual property liability, or contract breaches, potentially leading to legal action against individuals or the university.
- Reputation risks: Poor-quality research generated by AI tools can harm the university's reputation and make it harder to attract funding or research partners.
- Reputation risks: Multiple system and data breaches may deter partners from sharing their data and resources with the university.
Did you know that UTS Library Research Assistant, and Dimensions have AI assistants to help you find research papers and can be accessed with your UTS login?