As researchers in Human-Computer interaction, we have made a groundbreaking discovery – the placebo effect can extend to adaptive AI (artificial intelligence) interfaces! In our latest experiments, we asked participants to solve word puzzles while being supported by either no system or an adaptive AI interface. However, all participants experienced the same level of difficulty and did not receive any actual support from an AI.
Despite this, our results showed that the belief of receiving adaptive AI support increased participants‘ expectations regarding their own performance, which remained even after the interaction ended. These expectations were positively correlated with performance, as indicated by the number of puzzles solved.
Our findings have important implications for the assessment of AI-based user interfaces and other novel technologies. We believe that the descriptions of these systems can elicit placebo effects through the manipulation of user expectations, which could potentially bias the results of user-centered studies.
Overall, this research sheds new light on the power of the placebo effect and the role of user expectations in the evaluation of technology.
Read more here: https://dl.acm.org/doi/abs/10.1145/3529225