The Type of Task Matters: Conformity Towards Humans Versus AI in Different Task Domains

The rapid advancement of generative artificial intelligence (GenAI) has introduced new possibilities for human-AI interaction, with recent research highlighting the potential of AI to influence human decision-making in various domains. Few studies have examined how different task types affect human conformity to AI versus human influences. The current study investigates the impact of task context on informational conformity to AI compared to humans while also exploring the roles of perceived task objectivity and prior GenAI usage. We engaged 245 participants in six diverse tasks: morality judgments, opinion formation, knowledge questions, abstract reasoning, dot counting, and aesthetic evaluations, with each task presenting purported AI or human responses. Our results reveal that participants showed greater conformity to AI in the analytical tasks of dot counting and certain knowledge questions while conforming more to humans in tasks involving moral judgments, opinion formation, and aesthetic evaluations. Interestingly, neither perceived objectiveness nor prior GenAI usage consistently predicted conformity behavior across tasks. These findings suggest that the nature of the task significantly influences human-AI conformity patterns.