How humans engage with algorithms when interactions affect others

Determine how individuals engage with generative artificial intelligence during social interactions that entail consequences for other people, including how they respond when algorithms make decisions affecting others.

Background

The paper examines human responses to the use of generative AI (ChatGPT) in two-player economic games where decisions have real payoff consequences for both parties. Prior to this study, the extent and manner of human engagement with algorithms in such consequential social settings had not been clearly established. The authors highlight this uncertainty as a motivating gap for their experimental investigation.

References

In particular, it remains unclear how individuals engage with algorithms when the interaction entails consequences for other people.

— Generative AI Triggers Welfare-Reducing Decisions in Humans  (2401.12773 - Dvorak et al., 2024) in Abstract

How does interacting with the AI affect the mediator's ongoing interaction with the intended recipient and the broader social context in which the guidance is delivered? The present study provides initial observations concerning (a) and (b) within a single domain and offers design observations relevant to (c). Question (d) emerges from our design analysis but is not evaluated in the present preliminary study.

— Designing Human-mediated AI Guidance: Ready Together for Personalized Family Emergency Preparedness  (2608.19950 - Kurashvili et al., 20 Aug 2026) in Section Introduction