Giesen and Rothermund (2022) found that observationally acquired stimulus-response binding and retrieval (oSRBR) effects only occurred if participants believed to be interacting with another human, while effects were absent when they believed that they were interacting with a computer. However, in further unpublished studies from our lab, this finding could not be replicated consistently, suggesting that there might be other factors that influence whether people rely on observed actions performed by AIs/computers to guide their own actions. In two experiments, we investigated whether the expectation to be interacting with another human vs. an AI modulated oSRBR effects. To manipulate expectations, at the start of the experiment, participants were either told that they were going to interact with another human or with an AI. However, after a brief connection attempt, participants in the human condition were told that there were no other participants available, and that they would therefore do the experiment with an AI. Then they performed an online interactive color classification task to assess oSRBR effects. We expected to only find oSRBR effects if participants expected to interact with an AI from the start, but not if their initial expectation to be connected to another human was frustrated. Contrary to our expectations, we did no find a modulation of oSRBR effects by expectations in experiment 1. However, due to a significant number of drop-outs in the condition, in which a human partner was expected, we are currently replicating the study. Results will be presented and discussed.