Agentic AI and the carbon footprint of everyday activities

Our lives and interactions with the external world are being transformed by artificial intelligence (AI). The increasing role of artificial intelligence (AI) in decision-making has prompted discussions about individuals' perceptions of AI in terms of agency and control. This study examines individuals' willingness to delegate decision-making about carbon-intensive behaviours to an agentic AI system, and how the satisfaction of basic psychological needs influences its acceptance. ‘Agentic AI’ is defined as “an autonomous system that can make decisions and execute tasks without human intervention”. If individuals are willing to delegate decisions to agentic AI, especially in high AI-agency conditions, AI could optimise resource consumption by making data-driven, low-carbon choices. ‘High AI-agency conditions’ are defined as “full delegation of decision making and task execution to an autonomous AI system”. ‘Low AI-agency conditions’ are those in which users retain decision making functions, informed by AI recommendations. Self-determination theory (SDT) is a macro theory of motivation. It explores the conditions under which people feel motivated and engaged to perform a task successfully (Ryan & Deci, 2000). SDT identifies three basic psychological needs - competence, relatedness, autonomy - that, if satisfied, allow personal growth, enjoyment of effort, and overall well-being. In SDT, competence refers to an individual’s perception of their ability to perform a task effectively and achieve a desired outcome, relatedness refers to a person’s experience with caring for and connecting with other people, and autonomy refers to a person’s experience with free decision-making power over their own actions(Bergdahl et al., 2023). This dataset was created to explore how SDT constructs influence attitudes and intentions to use agentic AI systems to delegate energy-related decisions and how environmental concerns and AI literacy interact in the acceptance of agentic AI.