Guiding the Release of Safer E2E Conversational AI through Value Sensitive Design
Guiding the Release of Safer E2E Conversational AI through Value Sensitive Design
复制标题
DOI：
10.18653/v1/2022.sigdial-1.4
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
A. S. Bergman;Gavin Abercrombie;Shannon L. Spruit;Dirk Hovy;Emily Dinan;Y-Lan Boureau;Verena Rieser
中科院分区：
文献类型：
--
作者：
A. S. Bergman;Gavin Abercrombie;Shannon L. Spruit;Dirk Hovy;Emily Dinan;Y-Lan Boureau;Verena Rieser
在过去的几年里，端到端的神经对话代理极大地提高了它们与人类进行无限制、开放领域对话的能力。然而，这些模型通常是在来自互联网的大数据集上进行训练的，因此，可能会从这些数据中学习不良行为，如有毒或有害的语言。因此，研究人员必须努力解决如何以及何时发布这些模型。在这篇文章中，我们综述了最近和相关的工作，以突出价值观之间的紧张关系，潜在的积极影响，以及潜在的危害。我们还提供了一个框架，以支持从业者决定是否以及如何发布这些模型，遵循价值敏感型设计的原则。
Over the last several years, end-to-end neural conversational agents have vastly improved their ability to carry unrestricted, open-domain conversations with humans. However, these models are often trained on large datasets from the Internet and, as a result, may learn undesirable behaviours from this data, such as toxic or otherwise harmful language. Thus, researchers must wrestle with how and when to release these models. In this paper, we survey recent and related work to highlight tensions between values, potential positive impact, and potential harms. We also provide a framework to support practitioners in deciding whether and how to release these models, following the tenets of value-sensitive design.