Increasing the effectiveness of charitable giving using human-AI dialogues

Charitable donations frequently fail to maximize cost-effectiveness (the amount of good a donation does per dollar). This failure is often attributed to charitable motivations being affective and thus insensitive to evidence-based arguments. We challenge this perspective, hypothesizing that evidence can substantially increase effective giving—if that evidence is sufficiently compelling. We test this prediction in a pre-registered experiment (N = 1,949 Americans) by leveraging the ability of artificial intelligence large language models (LLMs) to engage in evidence-based back-and-forth dialogues. Participants allocated $1 between their favorite charity and a highly effective charity (the Against Malaria Foundation), before and after a conversation with an LLM advocating for the effective charity, a static LLM-generated persuasive message, or a control conversation. The LLM conversationsignificantly increased effective donations (45.9%), a significantly larger increase than the static message or control, as well as shifting moral attitudes. Effective giving can be meaningfully increased through evidence-based dialogues.