AI and altruism - An experiment on joint decision-making

Sammanfattning

Humans increasingly interact with artificial intelligence (AI) in various contexts. Through a pre-registered experimental survey with university students (N=115), I investigate how charitable giving varies between people who make a joint decision with an imagined AI agent compared to those who make the decision with an imagined human. I also investigate whether people who decide with an imagined AI tend to revise their share donated to a different degree in an individual decision, compared to those who make the same decision with an imagined human. I do not find any statistically significant support for the hypothesis that people who make the decision with an imagined AI, on average, give a different amount to charity or change their share of endowment given compared to those who make the decision with an imagined human. The ability to draw conclusions from this finding is, however, limited by the aspect that the co-player is hypothetical and that I have a relatively small sample.