What's Happening?
Field epidemiology training programs (FETPs) are grappling with the increasing use of generative artificial intelligence (AI) by trainees for tasks such as drafting code, text, and analyses. A survey revealed that two-thirds of fellows in Canadian, U.S.,
and European FETPs use AI in their work, primarily for analytic code, yet only one in five had received AI training. Concerns have arisen from evaluations showing AI tools providing inconsistent answers, citing non-existent guidance, and misreporting measures of association. In response, experts propose that assessment in FETPs should evaluate both the trainee’s reasoning and the quality of the AI-assisted product. This includes linking AI permissions to demonstrated competence, requiring records of substantive AI assistance, pairing major deliverables with oral assessments, and incorporating review and escalation criteria in outbreak workplans. The goal is to ensure independent competence while leveraging AI as a tool.
Why It's Important?
The integration of AI into field epidemiology training has significant implications for public health in the U.S. and globally. While AI can enhance efficiency, the potential for errors, misinterpretations, and over-reliance on AI tools poses risks to the accuracy and reliability of epidemiological investigations. Strengthening competence assessment ensures that future epidemiologists possess the critical thinking and analytical skills necessary to validate AI outputs and make informed decisions, especially during public health crises. This is vital for maintaining public trust in health data and interventions. Without proper assessment, there is a risk of producing epidemiologists who may lack the foundational understanding to identify and correct AI-generated inaccuracies, potentially leading to flawed public health recommendations and responses.
What's Next?
FETPs are advised to implement five key arrangements: linking AI permissions to demonstrated competence, requiring proportionate records of AI assistance for major deliverables, pairing major deliverables with oral assessment and observed independent performance, including review and escalation criteria in outbreak workplans, and evaluating these arrangements before wider adoption. These measures aim to ensure that trainees develop independent reasoning skills while responsibly utilizing AI. Programs will need to prepare mentors to assess AI-assisted work and ensure institutional support for approved tools and data governance. The evaluation of these new assessment strategies will be crucial to refine practices and ensure that AI effectively supports, rather than undermines, the development of competent field epidemiologists.
Beyond the Headlines
The debate over AI use in field epidemiology training touches upon broader ethical and pedagogical challenges in professional education. It highlights the tension between leveraging advanced technology for efficiency and preserving fundamental human skills like critical thinking, data interpretation, and ethical reasoning. The proposed assessment framework emphasizes human accountability and the need for trainees to understand the 'why' behind AI-generated results, not just the 'what.' This approach could serve as a model for other professional fields integrating AI, ensuring that technology augments human expertise rather than diminishing it. It also raises questions about equitable access to AI tools and training, particularly for trainees from diverse linguistic backgrounds, and the need for AI systems to be free from biases that could impact public health outcomes.













