This study explores whether people perceive phantom costs when interacting with robots, similar to how they perceive phantom costs when interacting with humans making overly generous offers. The key findings are:
Replicating previous research, the study found that people perceive phantom costs and are less likely to accept overly generous offers (a cookie plus $2) from human agents compared to a simple cookie offer.
The study found that people also perceive phantom costs and are less likely to accept overly generous offers from robot agents, across both screen-embodied and physically-embodied conditions.
While people were more likely to accept offers from robots compared to humans overall, the money backfire effect (being less likely to accept the offer with the additional $2) was observed for both human and robot agents.
The embodiment of the agent (screen vs. physical) influenced the likelihood of accepting offers from human agents, but not from robot agents. Physically embodied human agents were more likely to have their offers accepted compared to screen-embodied human agents.
Participants cited various criteria for their decisions, with phantom costs being a key factor, especially for the overly generous offers. Other factors included excitement, hunger, perceptions of the agent, sanitation concerns, and social etiquette.
The study provides new insights into how people make decisions when interacting with robots, showing that they can perceive hidden intentions and phantom costs, similar to how they perceive such factors when interacting with humans. This has important implications for designing transparent and trustworthy robot behaviors.
เป็นภาษาอื่น
จากเนื้อหาต้นฉบับ
arxiv.org
ข้อมูลเชิงลึกที่สำคัญจาก
by Benjamin Leb... ที่ arxiv.org 04-12-2024
https://arxiv.org/pdf/2404.07409.pdfสอบถามเพิ่มเติม