Question for algo accountability researchers:

What's your favorite empirical paper that includes data on actual decisions by humans who are aided by an automated decision-tool like a pre-trial risk assessment algorithm or HR algorithm or medical screening tool?