“Let’s Change the Subject”: How Virtual Assistants Counter Sexual Harassment
Amanda Cercas Curry Interaction Lab Heriot-Watt University Edinburgh, UK \AndVerena Rieser Interaction Lab Heriot-Watt University Edinburgh, UK
Abstract
How should conversational agents respond to verbal abuse through the user? To answer this question, we conduct a large-scale crowd-sourced evaluation of abuse response strategies employed by current state-of-the-art systems. Our results show that some strategies, such as “polite refusal” score highly across the board, while for other strategies demographic factors, such as age, as well as the severity of the preceding abuse influence the user’s perception of which response is appropriate. In addition, we find that most data-driven models lag behind rule-based or commercial systems in terms of their perceived appropriateness.
中文速览
当对话系统（如智能音箱、聊天机器人）遭遇用户的言语骚扰时，它该如何回应才算恰当？研究者收集了与真实用户的60万条对话，提炼出109条典型骚扰话语，再从Alexa、Siri等商业系统和多种数据驱动模型中采集共2441条回应，通过大规模众包评测让472名用户打分。结果显示，"礼貌拒绝"在所有人群中获评最高，而用户的年龄和骚扰类型会显著影响对其他策略的接受度——比如年龄较大的用户对"开玩笑式回应"非常反感，针对性骚扰要求时"回避"策略最受认可。更值得警惕的是，多数数据驱动模型的表现甚至不如规则系统，其中用"干净数据"训练的Seq2Seq模型反而频繁生成被认为带有调情意味的回复，与成人聊天机器人不相上下。这项研究为设计能随情境和用户特征自适应调整的骚扰应对策略提供了首批系统性实证依据，对构建更负责任的对话AI具有直接参考价值。
原文 arXiv:1909.04387;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1909.04387v1