Insight: Bias and Noise in Humans & AI: When to Trust Humans & Machines in Decision-Making .
Vol. 20, No. 4, 2022
When should we trust machine-based and human decisions in finance? In this article I answer this question by drawing on two sets of insights about decision error. I first draw on research of leading theorists on human decision-making and prediction, summarized through a set of articles and conversations with them about the two sources of decision error, namely, bias and noise. I also draw on two decades of experience operating a machine-learning based trading platform, where algorithmic bias and noise also manifest themselves, but very differently than in human decision-making. This two-pronged analysis of the properties of humans and algorithmic decision-making provides a backdrop against which the challenges and opportunities for creating trustable decision-making systems in finance come into sharp focus.