Artificial Intelligence and Public Health

Abstract Artificial intelligence (AI) is fundamentally changing public health, health surveillance and epidemiology. In last year’s International Festival of Public Health, a challenge for delegates was to consider if this is the “beginning of the end”? In a mere 10 months, we have seen an exponential rise in the use of AI across all disciplines. We will be exploring the opportunities AI brings to health improvement, health protection, healthcare public health and the research that underpins policy and practice. AI could help us tackle inequalities, digital exclusion and be that vehicle for bringing people out of poverty, but at what cost? The promise of AI to make healthcare “more productive”, potentially exacerbating inequities that stem from decades of underinvestment in health systems and health information. As budgets are tightening, there might be temptation to apply AI to health with the promise of efficiency and precision, pushing public health to grapple with the narrative of who is productive versus who needs support, and where resources are channelled for public health. AI-driven epidemiological analyses could also exacerbate the use of demand indicators, moving us further away from public health teams understanding the needs of their populations, especially the most vulnerable. This year, OpenAI released their report “AI as a Healthcare Ally” outlining the scale to which ChatGPT users ask questions about their health. They report that there are 40 million users asking about health every day, 200 million asking about health-related matters at least once a week, with 55% of those surveyed using AI to “check or explore symptoms” and 44% to “learn about treatment options”. Not only does this raise questions around how we ensure that people are equipped to understand the information they receive, but we must also address the issue of who people trust to provide them with health-related information. AI is fundamentally changing health communication, evolving from a tool for automation into a critical “communication infrastructure”. Yet, for public health, accuracy alone is insufficient. To be effective, information must be culturally resonant and earn the trust of the community it serves. Current evidence highlights a concern: AI often performs inconsistently across different languages and cultural contexts, which risks widening the gap for communities already facing healthcare barriers. Even a technically “perfect” AI response can fail if it ignores the patient’s lived experience or cultural nuances. Moreover, the role of bias in AI is well documented and presents significant concerns for equity. As these tools become embedded in healthcare, the challenge is no longer simply whether AI provides correct information, but whether people actually understand it, trust it, and feel that it speaks to their own social and cultural realities. The implications of AI for health are wider than only its performance. Water scarcity is a serious public health issue, yet AI datacentres consume huge volumes of water for cooling. We are contending with delivering equitable net zero transitions and how to provide clean affordable energy, while simultaneously building new AI infrastructure which will place significant demand on energy systems. However, AI also presents opportunities; the boom in AI-related construction and adoption of technology does present a route to improve health through employment (which must be balanced against its risks through job losses), training and skills development in the new technology era. We need a conversation about AI and the essence of core public health policy, practice and research as a social justice issue, where we leave no one behind and focus on the needs of the communities we serve.