The document provides a detailed overview of a September 16, 2024, webinar organized by COGR, focusing on innovative strategies for enhancing research administration efficiency through artificial intelligence (AI). The event featured a range of speakers from prominent institutions discussing the practical application of AI tools, including robotic process automation (RPA), LLM-based workflows, and chatbots, in the context of research administration. It opened with housekeeping announcements and guidelines on how to participate in the webinar, followed by a presentation of poll results revealing that the majority of participants’ institutions are not yet using AI, with key barriers including lack of knowledge, perceived cost, privacy concerns, and a cautious, observational approach.
The core of the webinar explored both theoretical and practical aspects of implementing AI in research administration. Presentations highlighted how AI and automation can streamline administrative workflows, improve accuracy and efficiency, and potentially reallocate human resources to higher-level tasks. Real-world use cases included automating IRB protocol data entry and categorization at the University of Texas at Arlington, and using bots to process subawards at the Mayo Clinic, resulting in significant time and labor savings. The Mayo Clinic also described the deployment of AI-driven contract review and negotiation tools. Additionally, the discussion covered the integration of AI-powered chatbots in research administration, emphasizing their ability to provide 24/7 support, automate repetitive queries, and facilitate stakeholder engagement. Throughout, speakers addressed challenges such as technical integration, training AI for specific institutional requirements, maintaining data privacy, managing costs, and overseeing AI accuracy. Despite these hurdles, the overall tone underscored the promise of AI to advance research administration, advocating for a thoughtful, phased adoption tailored to institutional readiness and priorities.
