Researchers

Use of Generative AI for Researchers

The contribution of artificial intelligence technologies to the research data lifecycle is undeniable. Data collection, cleansing or analysis that could previously take months can now be completed in hours thanks to these technologies.

Data Lifecycle

Figure 1. Data Lifecycle Chart (Source: Harvard Medical School)

Basic Principles

• Researchers should identify the scope and risks of the GAI programs they intend to use in advance. These risks include serious legal issues such as data security and copyright.
• Project Managers (PMs) must raise awareness among project staff about the ethical and legal use of artificial intelligence. Koç University’s Research Data Policies assign PMs a high level of responsibility in this regard.
• It is important that GAI technologies to be used in research projects receive technical approval from the Koç University Information Technologies (IT) Directorate. This process can be managed by the Research Data Management Group on behalf of the researcher.
• The IT Directorate may put restrictions on the use of GAI technologies with information security risks in Koç University infrastructure. However, for the progress of research projects, IT specialists can develop alternative solutions together with the researcher.

Please visit the KU Research Data Management Group page for more information

Data Management Planning and Artificial Intelligence

An increasing number of sponsoring organizations are requesting data management plans for research projects. In these plans, some of the key issues about AI that are important to cover are as follows:

• Are artificial intelligence technologies being used at any stage of the research data lifecycle?
  • Will research data be generated with open-source GAI technologies?
  • Will data be retrieved from social media APIs using artificial intelligence technologies?
  • Are statistical analyses being conducted on research data using open-source GAI technologies?
• Is a new artificial intelligence technology being created as a result of the research project?
• Are data that may contain personal information being processed using open-source GAI technologies?
• Is it necessary to apply to ethics committees for artificial intelligence technologies to be used or produced during the research project?

KU researchers who answer “yes” to at least one of the above questions should contact the Research Data Management Group via (i) rdm@ku.edu.tr or (ii) track-it service.

Data Creation and Artificial Intelligence

Research data collected through open-source GAI programs come with certain risks. It’s essential for researchers to approach these risks thoughtfully and stay within legal boundaries:

• License Infringement: Like any commercial product, research data is subject to usage licenses. It is unclear under which license GAI outputs are collated from shared data. Therefore, GAI outputs carry the risk of license infringement.
• Inaccurate Data Generation: GAI technologies can produce hallucinations when trying to generate information from sources with limited access. The researcher should apply serious quality control to the outputs of GAI.
• Ethical Concerns: GAI programs may lack a consistent ethical filter. Researchers must ensure that GAI-generated outputs comply with relevant ethical guidelines. In particular, outputs that may contain biases can lead to serious ethical issues.

Data Sharing and Artificial Intelligence

Researchers should avoid analyzing data containing personal information using open-source GAI programs.

• All information processed by GAI programs contributes to the algorithms behind them. This means personal information processed by GAI could potentially be exposed as open data.
• Researchers should carefully anonymize or pseudonymize personal data before using GAI. Additionally, they should be aware of the limitations of anonymization techniques and consider alternative approaches, such as synthetic data generation, to protect privacy.
• In case anonymized data is to be processed using open-source artificial intelligence programs, it is important for the Project Manager (PM) to contact the Data Privacy and Research Data Coordinators.

Data Sharing and Artificial Intelligence

If artificial intelligence technologies were used during the collection and processing phase, the researcher should provide appropriate citations.