Explaining the inexplicable: A study of people's reactions to futuristic AI

Abstract

To better understand people's reactions to futuristic and seemingly impossible technologies of Artificial Intelligence (AI), we studied the responses of 21 students to a phone app that performs two highly surprising conjuring tricks simulated through a Wizard of Oz method: uncannily accurate lie detection, and control over a user's apparently free choices. We observed how, in the absence of readily available interpretations, our participants generated a range of pseudo explanations for how the AI performed its feats, which we classified as non-explanation, unresolved, concordant, and revelatory. We interpret these findings in terms of Daniel Dennett's notion of explanatory stances, and sugge..