The author further criticizes the medical system for often dismissing patient experiences and expertise, leading to delayed diagnoses and incorrect data. They argue that these issues will propagate into AI unless researchers include meaningful patient participation. The author concludes by advocating for a clear-eyed assessment of the risks and opportunities of AI in medicine, and for the development of participatory approaches to machine learning and patient-led medical research.
Key takeaways:
- AI has made significant strides in the medical field, but there are concerns about its practical impact on patients, particularly in terms of disregarding patient perspectives and exacerbating inequality.
- Automated decision-making systems are increasingly used in areas that significantly impact people's lives, such as jobs, housing, and healthcare. However, these systems often contain errors and are designed to prioritize corporate and government revenues over the needs of the poor.
- The medical system often disregards patient experiences and expertise, leading to delayed diagnoses, misdiagnoses, missing data, and incorrect data. This issue is not something AI can solve, as it is great at finding patterns in existing data but cannot address the problem of missing and erroneous underlying data.
- While AI holds transformative possibilities for medicine, it is important to be clear-eyed about both the risks and the opportunities. Ignoring the realities of the medical system will lead to the design of AI for an idealized medical system that does not exist, potentially leading to further disempowerment of patients.