NTU Position on the Use of Generative Artificial Intelligence in Research

Nanyang Technological University (NTU) recognizes that Generative Artificial Intelligence (GAI) – such as OpenAI’s ChatGPT, Microsoft Bing, and Google Bard – provides new tools to support research practices and scholarly activities, including the preparation and drafting of research proposals, manuscripts, and other scholarly work.

NTU acknowledges that GAI has the potential to enhance the quality and efficiency of research, and provide new modes of inquiry. However, it is important to recognize that its use in research can raise ethical, research integrity and other concerns. Researchers must therefore carefully evaluate these issues, and have a thorough understanding of the technology – including its potential risks and limitations (e.g., current AI-generated outputs can contain bias, errors, inaccuracies, or falsehoods) – before considering its use.

Researchers who use AI tools in their scholarly work must use GAI in a responsible and accountable manner, and be transparent on the extent and nature of the involvement of GAI in their work. This would include acknowledging the use of any AI tools in their research proposals, manuscripts, and scholarly works.

Currently, NTU does not impose restrictions on the use of GAI in research, except in the scenarios outlined in points 3 and 4 below.  

The Responsible Use of GAI in Research

1.  Acknowledging/Declaring the Use of GAI

In the interests of transparency and integrity, the use of GAI of beyond basic spelling and grammar checks should be appropriately acknowledged and cited.

Additionally, any use of GAI to generate images for illustrations or figures should be clearly stated in the caption and/or document, in a manner easily understandable to the reader.

2.  Authorship

Authorship requires the acceptance of responsibility for the work described in any manuscript. Researchers should recognise that GAI cannot be held responsible as an author for the accuracy, integrity, and content of such work.

Therefore, any GAI (e.g. ChatGPT) will not be listed as an author of any paper with an affiliation to NTU; or listed as a Principal Investigator (PI), Co-PI, or collaborator in any research proposals.

Authors and/or PIs are fully responsible for the content of their scholarly materials (e.g., research proposals, grant applications, manuscripts for publication) in which GAI was used in the preparation and/or development of such outputs. Researchers must therefore exercise caution and judgement when using GAI, and be ready to verify the accuracy and validity of their work. (See Figure 1 under resources.)

3.  Data Privacy & Confidentiality

The use of GAI to process or analyse research data must comply with all relevant data privacy and protection laws, regulations, and institutional policies - e.g., the Personal Data Protection Act (PDPA), NTU’s Data Governance Policy.

Any confidential or sensitive information, and/or personal data are not to be uploaded to any GAI software, system, or platform unless:

1. Access to the GAI is controlled and restricted to only authorized study members involved in the research;
2. The data is not retained in or by the GAI; and
3. The activity does not contravene any applicable laws, regulations, or institutional policies in the process.
   

Researchers must prioritize and safeguard the privacy and confidentiality of research data when using GAI and will be held responsible for any leakage of data.

4.  Data Owned by Non-NTU Entities

Data owned by external parties, including but not limited to businesses, organisations, and government ministries and agencies, are also not to be uploaded to any GAI software, system, or platform unless:

1. All requirements for confidential, sensitive data, and/or personal data listed under “Data Privacy & Confidentiality” are satisfied;
2. Valid written permission has been explicitly provided by the data owner/custodian; or
3. The use of such GAI has been agreed upon in the Research Collaboration Agreement (RCA).

5.  Rules established by other Parties

If the proposal, manuscript or other document is being submitted to a agency, journal or other party with rules concerning GAI that differ from the NTU position, then the stricter rules should be followed. NTU researchers may contact RIEO for advice.

In conclusion, Generative AI in research must be used in an ethical, responsible, transparent, and accountable manner that benefits society. NTU faculty, staff, and students should adhere to the following when employing GAI in their research:

1. Appropriately acknowledge and cite the GAI used;
2. Do not attribute or list any GAI as an author;
3. Take full responsibility for the use of any GAI-generated content in their work; and
4. Safeguard the privacy and confidentiality of confidential and/or sensitive research data, including Personal Data.

Date issued: 4 Jul 2023

Disclosure

OpenAI’s ChatGPT (Mar 23 version) was used to improve the clarity and readability of this statement. No confidential or sensitive information was uploaded during this process.

Resources

UNESCO (2023). ChatGPT and Artificial Intelligence in Higher Education: Quick start guide. 

Figure 1: When is it safe to use ChatGPT?

(UNESCO / ChatGPT and Artificial Intelligence in higher education: Quick start guide -

licensed under CC-BY-SA 3.0 IGO)