Abstract
AI systems in medicine and healthcare are being extensively explored in prevention, diagnosis, novel drug designs and after-care. The application of AI technology in healthcare systems promises impressive outcomes such as equalising healthcare, reducing mortality rate and human error, reducing medical costs, as well as reducing reliance on social services. In the light of the WHO "Decade of Healthy Ageing", AI applications are designed as digital innovations to support the quality of life for older persons. However, the emergence of evidence of different types of algorithmic bias in AI applications, ageism in the use of digital devices and platforms, as well as age bias in digital data suggests that the use of AI might have discriminatory effects on older population or even cause harm. This paper addresses the issue of age biases and age discrimination in AI applications in medicine and healthcare systems and try to identify main challenges in this area. It will reflect on the potential of AI applications to amplify the already existing health inequalities by discussing two levels where potential negative impact of AI on age inequalities might be observed. Firstly, we will address the technical level of age bias in algorithms and digital datasets (especially health data). Secondly, we will discuss the potential disparate outcomes of automatic decision-making systems (ADMs) used in healthcare on the older population. These examples will demonstrate, although only partially, how AI systems may create new structures of age inequalities and novel dimensions of exclusion in healthcare and medicine.