Under AI Watch: Understanding Online Behaviors under Supposed AI- Surveillance

Artificial Intelligence (AI) systems being capable of mimicking human intelligence to perform tasks have raised legitimate concerns around ethical and societal implications of their implementation. Despite the fast-paced reproduction of ethical principles to ensure safe and accountable deployment, it would be irrational to consider these sufficient. The adoption of these frameworks heavily relies on citizens’ acceptance to the content and the approach of AI implementation. This study focuses on evaluating citizens’ behaviours in reaction to assumed AI in online spaces, the factors that trigger rejection and potential changes in behaviour, including potential counteractions. Using an online experiment on Facebook, 30 participants were asked to perform eight tasks, accompanied by think-aloud methodology, under the assumption of AI-surveillance. The findings provide a detailed understanding of the types, reasons, and rationales for agreeing or disagreeing to conduct tasks under assumed AI-surveillance within reallife settings.