Search

As we introduce complex algorithmic systems into decision-making in high-stakes domains, system designers need principled approaches to help people set their expectations of these systems, and give them mechanisms for recovery when these systems fail. For example, a doctor using a machine learning-based system in clinical care needs to know when they can expect the model to perform well, and when they should not rely on its output. However, supporting these kinds of judgments is difficult when stakeholders have conflicting needs and goals, or cannot directly assess the quality of a system's output. This dissertation examines two such contexts: matching algorithms for assigning students to public schools; and machine translation systems, which use machine learning to translate between natural languages. I take three approaches to design for reliability in these contexts: first, teaching users what a system can and cannot do; second, aligning system evaluations with people's actual use cases, needs, and goals; and third, helping users recover from failures. By making it easier for users to understand what kinds of inputs are supported by a system, and how they can express their intent within that supported scope, we can increase users' agency to make appropriate decisions about how and when to use these systems in high-stakes settings.

Solid analytical work in philosophy has clearly defined the real possibility of existential harm to humanity by general artificial intelligence (AI) and argued that existential risk prevention is a global priority. Such existential risk scenarios of a “super AI” exterminating humankind have a rich and powerful attraction to our imagination. In contrast, here I claim that the main practical aim of the philosophy of AI for health should be the study of non-existential risks of AI health interventions, including AI scientific health claims. I will call this “the duty to study the non-existential risks of AI health interventions”. To support this duty, I will argue that harm to individuals, groups, or populations with AI health interventions is not just a probability that may happen in the future but is instead a reality already happening in the present, with consequences for our understanding of traditional duties and something that can be averted or minimized with simple safeguards. The Surgisphere database scandal during the COVID-19 pandemic illustrates the first point. The AI scientific claims —published in The Lancet and NEJM, based on a fraudulent global database that allegedly used AI/ML to automate processes— made governments and the World Health Organization (WHO) change their health policy with harmful results. If this argument is sound, three traditional duties of health professions gain a new meaning in the philosophy of AI for health. First, the duty to care brings support to the main idea of this presentation that the philosophy of AI for health should prioritize the study of non-existential risks, as we prioritize some patients to emergency care over research in medicine. Second, once we know of non-existential risks, the duty to do no harm justifies implementing appropriate safeguards in the work of health professionals (e.g. medicine, nursing, pharmacy, and public health) and data scientists who are currently developing and using AI health interventions. Third, the duty of justice states that we should redress unfair inequalities related to AI health interventions, such as improving AI education for health professionals and the national government's capacity to evaluate AI scientific claims —usually from low- and middle-income countries that are at a higher risk of non-existential harms. Finally, I will share some examples of simple appropriate safeguards (e.g., SPIRIT-AI guidelines for clinical trial presentations, support from WHO) and attempt to address general objections to its implementation such as futility, stifling the benefits of health innovation, and trade secret confidentiality, Lecture presented at the Symposium "Intelligent Oncology: The Potential of AI to Cure Cancer", Freiburg, Germany, 30.09.22. Acknowledgments: this work has been done under the visiting research scholarship of the Mertelsmann Foundation

The COVID-19 pandemic has highly impacted the communities globally by reprioritizing the means through which various societal sectors operate. Among these sectors, healthcare providers and medical workers have been impacted prominently due to the massive increase in demand for medical services under unprecedented circumstances. Hence, any tool that can help the compliance with social guidelines for COVID-19 spread prevention will have a positive impact on managing and controlling the virus outbreak and reducing the excessive burden on the healthcare system. This perspective article disseminates the perspectives of the authors regarding the use of novel biosensors and intelligent algorithms embodied in wearable IoMT frameworks for tackling this issue. We discuss how with the use of smart IoMT wearables certain biomarkers can be tracked for detection of COVID-19 in exposed individuals. We enumerate several machine learning algorithms which can be used to process a wide range of collected biomarkers for detecting (a) multiple symptoms of SARS-CoV-2 infection and (b) the dynamical likelihood of contracting the virus through interpersonal interaction. Eventually, we enunciate how a systematic use of smart wearable IoMT devices in various social sectors can intelligently help controlling the spread of COVID-19 in communities as they enter the reopening phase. We explain how this framework can benefit individuals and their medical correspondents by introducing Systems for Symptom Decoding (SSD), and how the use of this technology can be generalized on a societal level for the control of spread by introducing Systems for Spread Tracing (SST).