Amidst the escalating climate crisis and pressing environmental issues, fostering cooperative human behavior is indispensable for the realization of a sustainable society. While findings in environmental psychology have demonstrated that interpersonal communication can promote such cooperation, this study investigates the efficacy of communication with artificial intelligence (AI) as a practical application of this principle. To investigate this, both humans and AIs played the "Online waste management game," which we developed by adapting the iterative public goods game.
An online experiment was conducted with a total of 330 three-person groups under four conditions: three humans, two humans and one AI, one human and two AIs, or three AIs. In the game, participants decided how much of their allocated 20kg of waste to recycle. Recycling incurred point costs, but the total recycled amount was doubled and redistributed. The study included a control condition and a chat condition where participants could communicate anonymously after the fifth round.
Following the communication intervention, the rate of cooperation in all-human groups increased by an average of 10%. In contrast, groups with both humans and AIs saw an average increase of 20%. The analysis of chat showed that the percentage of groups that reached an agreement to cooperate was higher in human-AI mixed groups than in all-human groups.
These findings suggest that communication with AI may have a greater effect on promoting cooperative behavior than communication among humans. Whereas communication between humans is often constrained by logistical challenges such as scheduling, the use of AI can overcome these practical implementation costs. This opens up new possibilities for AI-driven interventions to promote pro-environmental behavior.