Evidence of spillovers from (non)cooperative human-bot to human-human interactions

= 83,411 decisions by 4,171 participants told they are interacting with a bot or human), we demonstrate that human-bot interactions do spill over to human-human ones, and in two ways. First, interacting with a bot reduces cooperation not only during the initial interaction, but also reduces downstream cooperation toward a new human partner. Additionally, bots' (like human partners') behavior matters: interactions with bots playing tit-for-tat promote cooperation, and interactions with noncooperative bots reduce cooperation toward downstream human partners. The implementation of bots in previously human-only spaces alters human cooperation, not just toward bots, but toward other humans as well.