Generative AI is a tool, not unlike other tools available to researchers such as websites, electronic journals, and aggregated full-text databases provided by university libraries. The difference is that GenAI platforms have the ability to create new content based on that which has been input into its learning analysis. As with any tool, users should be familiar with the advantages, limitations, and shortcomings that AI platforms can apply to research activity.
Cornell University has created a thorough report outlining considerations for using AI in scholarship and research, titled Generative AI in Academic Research: Perspectives and Cultural Norms.
In this report four stages of the research cycle are identified, with each impacted by AI. They are:
- Research Conception & Execution - An internal stage whereby processes such as hypothesis generation, research ideation, literature review, and research infrastructure. Suggested practices include organizing and orientation toward growing base of information, drafting literature reviews with established guardrails against hallucinations, and iteration toward refining reviews for new contributions and identified gaps.
- Research Dissemination - AI tools can aid researchers in providing more flexible clarity and equitable dissemination of related work, while raising serious challenges to issues of authorship, inherent bias, peer-review, and rigor of methodology.
- Research Translation - AI tools will have an impact on inventorship, patentability, and copyright in a legal sense, in combination with the commercial impact of proprietary systems. Currently, patents and copyright legally applies only to human beings, but users must be familiar with using tools with respect to data privacy/protection, commercialization and general AI literacy.
- Research Funding & Compliance - AI tools have a productivity impact when synthesizing information for sponsors or donors, but with the added responsibility placed upon the Principal Investigator (PI) to ensure accuracy of output as well as protection of data and privacy.
With each advantage awarded to the researcher(s) through the use of Generative AI, so too, exist serious areas of caution and discretion throughout the research cycle. Full transparency of use, and ongoing familiarity or training in AI literacy is essential to guarding against potentially flawed, inaccurate,or non-compliant behavior against the fundamentals of ethical and responsible research practice.
