There are long-held concerns that introducing computers into decision-making can lead to human operators becoming complacent or over-reliant on their computer support. This is referred to as the control problem and is discussed in this paper, published in the Minds and Machines Journal for Artificial Intelligence, Philosophy and Cognitive Science.  

Authors argue that, except in certain special circumstances, algorithmic decision tools should not be used in high-stakes or safety-critical decisions unless the systems concerned are significantly “better than human” in the relevant domain or subdomain of decision-making.