Keith Dear argues that AI will change decision-making in the defence and security arena in four principal ways. First, by enabling ‘cognitive manoeuvre’ the use of predictive analytics to enable much earlier intervention. Second, by forcing humans to take themselves ‘out of the loop’ for decision-making by out-performing them in an increasing number of domains. Third, by providing advice that is correct, but difficult to explain. Fourth, in the short term, by driving unprecedented rigour into decision-making processes, forcing decision-makers to be much more explicit about the mental models on which they are basing decisions, to enable comparison with automated analytics.