The effect of message framing and timing on the acceptance of artificial intelligence's suggestion

AI helps us make decisions in various domains such as healthcare, finance or entertainment (e.g. Netflix, IBM Watson and etc.). However, people's trust and acceptance of AI are highly susceptible to when and how the suggestion is presented. This study examined the role of the message framing and timing on acceptance when the performance of AI is stated. The study employed a 2 (message timing: before vs. after decision) x 3 (message framing: no information vs. negative framing vs. positive framing) between-subjects experiment where participants were told to solve the specific problem with AI in different conditions. The results showed that participants perceived the suggestion of AI more reasonable and accepted it more when the performance is not stated than any information is provided and they perceived the suggestion of AI more reasonable when the message is presented before the decision is made. The theoretical and practical implications are discussed.