Reinforcement learning and the prevention of data catastrophes

Abstract

Companies incur immense losses due to employee neglect to save and back up data and failure to frequently update anti‐virus protections. This problem appears perplexing as such oversights are clearly neither in the organization’s nor in the employees’ best interest. We review the possible reasons for this phenomenon arising from studies of social dilemmas, unrealistic optimism, and reinforcement learning. We follow with three examples of “under‐saving” behavior. The results reveal that in all three cases computer users, novices and experts, feel that they do not save enough. This feeling is consistent with the reinforcement learning account. People think that they are less careful than they wish to be. The implications of this observation are discussed.

Citation

Yechiam, E., Haruvy, E. and Erev, I. (2002), "Reinforcement learning and the prevention of data catastrophes", Journal of Managerial Psychology, Vol. 17 No. 7, pp. 599-611. https://doi.org/10.1108/02683940210444058