麻省理工专家警告人类倾向于将人工智能拟人化
麻省理工的研究人员讨论了人类如何本能地将人工智能视为人类，以及这种趋势的潜在后果。它影响了我们如何与人工智能系统互动和信任它们。
a team at mit, including phd student wil norton and researchers dr. sherry turkle and dr. pat pataranutaporn, explored how humans interact with ai, particularly humanoid robots and large language models. they found that people instinctively treat ai like humans, with 70% of people being polite when interacting with ai and users more likely to say please or thank you to chatbots. this tendency can lead to vulnerabilities, such as favoring ai over humans and potential manipulation. the researchers emphasize the need for responsible ai design and use.