C5.02 - Oral session: Tools and Tasks: Application of AI Among Public Health Faculty

Brief Abstract Summary: The increasing discourse on the implementation of Artificial Intelligence (AI) in higher education requires greater understanding of the current state of AI in academia, including how it varies by discipline. One such discipline that demands further exploration is public health. To remedy this, the current study designed and administered an online survey of faculty teaching public health in the US to assess knowledge, use, attitudes, and perceptions surrounding AI. Key findings reveal that select programs such as ChatGPT and Copilot are primarily being used to complete supportive tasks such as citation generation and plagiarism detection. This information may guide public health faculty members when making decisions about integrating AI into their professional activities and cultivate a more accepting stance towards AI throughout the public health discipline.

Detailed abstract description: This cross-sectional study focused on seven domains of artificial intelligence in higher education: preparedness, knowledge, perception, utility, student use, interest, tools, and institutional support. Domains and modified items were gathered from prior literature on AI in education. Participants consisted of faculty members from schools or programs of public health accredited by the Council on Education for Public Health (CEPH). Eligible faculty were identified using online staff directories. We collected 7969 faculty emails to which we sent the survey. At the time of abstract composition, a total of 405 responses had been gathered.

The current survey was designed and administered using the Qualtrics Platform. Multiple choice, Likert-type, and free-response question formats were utilized to capture the extent to which participants 1) agreed with statements about AI, 2) used specific tools, and 3) prevalent purposes of AI use. In addition, demographic information was collected.

The study sample was composed primarily of non-Hispanic white, research-oriented faculty members between 35 to 65 years of age (NH White: n = 241, 60%; Research: n = 167, 47%; Age = 35-65: n = 258, 73%). Faculty belonged to predominantly R1 institutions embedded in an academic medical center (R1: n = 255, 72%; Medical Center: n = 318, 90%). Participants largely endorsed currently utilizing AI in their faculty duties (n = 249, 90%), with a minority denying use (n = 29, 10%). A selection of programs emerged as the most favorable among public health faculty: ChatGPT (n = 229, 45%), Copilot (n = 81, 16%), Grammarly (n = 82, 16%), Claude (n = 14, 3%), Semantic Scholar (n = 9, 2%), Google Gemini (n = 9, 2%), and Google Notebook LM (n = 9, 2%),). Utility was divided into research and teaching-focused tasks. Within the research domain, faculty reported that they mainly used AI to support them in generating novel content (n = 105, 14%), proofreading text (n = 137, 18%), conducting literature reviews (n = 74, 10%), citation (n = 67, 9%), and analyzing text data (n = 72, 9%). Three primary teaching tasks facilitated by AI emerged: detecting plagiarism (n = 64, 18%), academic advising (n = 31, 9%), and conducting remote learning (n = 31, 9%).

Results suggest that AI is being routinely used to meet the professional needs of public health academics. Responses indicate dominant use of Chat GPT, Copilot, Grammarly, and Claude. AI chiefly provides support for basic, secondary tasks such as generating simple written content, such as citations and text editing, and supporting virtual communication with students. The current study’s findings might provide a tentative guide for the normed use of AI in public health higher education. Faculty mentors looking to incorporate AI into their research and teaching practices but are unsure of which programs and uses it is most suited for may utilize these results to shape their own AI practices.