Limitations of Generative AI in Research Grant Development

While Generative AI (Gen AI) can improve efficiency, it they cannot replace the intellectual work, specialized knowledge, or compliance verification performed by a researcher or research administrator. Researchers are always responsible for the proposals they submit.

Limitations in Funding Research:

  • Data Gaps and Cutoffs: Gen AI may suggest expired opportunities or miss new niche calls due to knowledge cutoffs or limited access to real-time, exclusive data. They are not a replacement for specialized funding databases or the university's Research Office.
  • Potential to miss critical nuances: GenAI may struggle to grasp subtle details and unstated priorities when summarizing funding announcements.

Limitations in Proposal Development:

  • Hallucination: GenAI can generate inaccurate information, fabricated citations, or flawed methodologies. Researchers must verify every citation, factual claim, and technical detail.
  • Data Security and Confidentiality Concerns: Researchers should never input confidential, unpublished, or proprietary research data into commercially available tools. Using institutional, enterprise-level tools (like Gemini accessed via CCID) is recommended for sensitive work. For more on data security, refer to the University of Alberta's Artificial Intelligence Data Safety Guidelines.
  • Plagiarism and Intellectual Property (IP) Violations: GenAI may generate text that closely resembles existing sources. Submitting proposals with such material can constitute plagiarism.
  • Financial and Budgetary Accuracy: GenAI can draft budget justifications but cannot perform accurate financial planning or integrate proprietary institutional financial data. Researchers are responsible for validating all numerical and financial information.

Sources: