How to Make an AI

Abstract

Abstract This chapter describes how health artificial intelligence (AI) development is embedded in multifaceted sociotechnical systems that involve not only cutting-edge advances in computer science and medicine, but also scientific standards, regulatory structures, and economic systems. It describes how algorithms find patterns in numeric representations, and how AI accuracy is measured following one of several conflicting approaches established by federal regulation and scientific communities. The chapter details how AI development is a complex choreography of humans and technology. It also offers an orientation to the technical details of AI development and explores how the many systems that compose AI are assembled into a single package or product. Along the way, the chapter describes how bias can enter the sociotechnical systems of deep medicine in different ways at different moments in the AI development process.

Ähnliche Arbeiten