The AI health divide: Three challenges and recommendations to improve design, transparency, and access for low-SES communities

Background Low socioeconomic status (SES) communities have been historically disadvantaged by food deserts, limited health and digital literacy, and restricted access to healthcare and digital technologies. These factors have increased the prevalence of chronic disease and contributed to reduced engagement with healthcare systems. Alongside this, the rapid increase in popularity of Artificial Intelligence (AI), integration into healthcare systems such as patient portals, symptom checkers, and care coordination platforms has become commonplace. However, the incorporation of AI assumes inflated levels of digital access, literacy, and cognitive capacity, which risks worsening existing inequalities. Objective This paper examines how AI-enabled digital integration into healthcare systems may inadvertently disadvantage low SES communities, while widening access disparities, exposure, and trust. It then proposes targeted recommendations to improve future AI design and implementation efforts. Methods This paper synthesizes evidence regarding social determinants of health, digital health literacy, and trust in AI to develop a set of practical design guidelines that support the needs of low-SES individuals. We draw from literature pertaining to human factors, health equity, digital health, and inclusive design. Results The analysis identifies barriers in which current AI tools and features misalign with low-SES individuals. We contextualize infrastructure gaps, high cognitive and literacy demands, individual differences, and the use behaviors related to the intersection of SES and AI in healthcare. Conclusion This paper presents design and implementation guidelines to reduce cognitive load, enhance accessibility, and build trust, while ensuring that AI integration in healthcare technologies supports rather than undermines health equity for low SES communities.