Artificial Intelligence and Ethics - Privacy & Security in the Health System

Betserai Albert Magarire

Since AI is a new technology that has recently been introduced to the world, many people still doubt whether we can trust the accuracy of AI. Two main aspects of medical diagnostic models are affected: the training and usage of the diagnostic models and the models’ access to patient data. There have also been concerns about the potential security breaches that can result in the abuse of patient data if the systems are not well monitored. Some of the issues that are raised emerge from the way the Artificial Intelligence models are trained. The devices in use to diagnose and treat patients have shown a tendency of being biased based on the racial differences, skin color differences, and genetic differences in the patients. This results in other patients getting inaccurate diagnoses. The main type of bias is the trifecta of human biases that are involved in supervised machine learning models and causes skewed training. This research seeks to answer the following question: To what extent should AI play a role in the diagnosis of illnesses?

• Betserai Albert Magarire graduated from Mazowe High School in Zimbabwe and is a computer information systems (engineering) major with a minor in cybersecurity. He has participated and received awards in multiple online hackathons and computer science summits. His remarkable projects include developing a learning management system for his high school and a proprietary POS System.