SDS 484: Algorithm Aversion — with Jon Krohn

Algorithm Aversion

Algorithm aversion is a cognitive bias where individuals prefer human forecasts over more accurate algorithmic predictions. Jon Krohn explains that this bias persists even when algorithms outperform human forecasters, particularly after witnessing both make similar mistakes. This aversion can lead to costly decisions, as people lose confidence in algorithms faster than in humans when errors occur 1.

Despite this, people are susceptible to an unfortunate cognitive bias called algorithm aversion, which is a costly preference for a forecast from a human instead of from a higher accuracy forecast by a statistical model or a machine learning model.

--- Jon Krohn

To counter this bias, Jon suggests validating algorithm performance with data and educating skeptical clients about this common, yet unfounded, aversion 1.

   

Study Insights

A pivotal study by Berkeley, Dietvorst, and colleagues at the University of Pennsylvania highlights the phenomenon of algorithm aversion. Jon Krohn notes that the study found people lose confidence in algorithmic forecasts more quickly than in human forecasts when both make the same error 1. This research underscores the irrationality of algorithm aversion, as it persists despite evidence of superior algorithmic performance.

In research published in 2015 by Berkeley, Dietvorst and his colleagues at the University of Pennsylvania, they observed that this algorithm aversion is caused by people losing confidence more quickly in an algorithm forecaster relative to a human forecaster when the algorithm and the human make the same mistake.

--- Jon Krohn

Jon encourages listeners to explore this study further, as understanding the bias can help in overcoming it and making more informed decisions 1.