FAI: A Human-Centered Approach to Developing Accessible and Reliable Machine Translation

这个人工智能公平项目旨在开发技术，以可靠地增强高风险环境下的跨语言交流，例如当一个人需要与不讲自己语言的人交流以获得医疗建议或申请工作时。虽然在这些情况下经常使用机器翻译技术，但现有的系统经常会出现错误，这可能会对患者或求职者造成严重后果。此外，当人们不理解翻译的源语言或目标语言时，要知道自动翻译在什么时候可能是错误的，这是一项挑战。这个项目通过为外行用户开发可访问和可靠的机器翻译来解决这个问题。它将提供引导用户识别和修复翻译错误的机制，并帮助他们在翻译不完美的情况下做出更好的决策。因此，更多的人将能够可靠地使用机器翻译来跨越语言障碍进行交流，这将对他们的生活产生深远的积极影响。具体来说，这个项目为机器人的自然语言处理和交互设计做出了贡献，机器人可以添加到任何基于文本的对话中，在这些对话中，它可以扮演类似于口译员的角色。机器人将引导用户为机器翻译编写适当的输入，帮助用户理解输出，并在检测到沟通不畅和会话中断时进行干预。机器人的设计将遵循以人为本的设计过程，包括需求发现研究、迭代系统开发和部署，以及通过控制实验进行的用户评估。在后端，机器人将依赖于自动检测翻译错误的质量评估模型，为最终用户提供有用的指导。该项目产生的数据、模型和设计建议将以多种方式推进计算研究。这将导致新的机器翻译质量评估技术，考虑到错误对最终用户的影响；它将扩大可解释人工智能研究的范围，以涵盖语言生成工具造成的相当大的风险和危害，它将产生新的界面设计，帮助外行用户理解人工智能系统。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This Fairness in AI project aims to develop technology to reliably enhance cross-lingual communication in high-stakes contexts, such as when a person needs to communicate with someone who does not speak their language to get health care advice or apply for a job. While machine translation technology is frequently used in these conditions, existing systems often make errors that can have severe consequences for a patient or a job applicant. Further, it is challenging for people to know when automatic translations might be wrong when they do not understand the source or target language for translation. This project addresses this issue by developing accessible and reliable machine translation for lay users. It will provide mechanisms to guide users to recognize and recover from translation errors, and help them make better decisions given imperfect translations. As a result, more people will be able to use machine translation reliably to communicate across language barriers, which can have far-reaching positive consequences on their lives.Specifically, this project contributes advances in natural language processing and interaction design for a bot that can be added to any text-based conversation, where it can play a role similar to an interpreter. The bot will guide users to write appropriate inputs for machine translation, help users understand outputs, and intervene when it detects miscommunication and conversational breakdowns. The design of the bot will follow a human-centered design process, consisting of need-finding studies, iterative system development and deployment, and user evaluations via controlled experiments. On the back-end, the bot will rely on quality estimation models that automatically detect translation errors to produce useful guidance for end-users. The data, models, and design recommendations generated by this project will advance computational research in multiple ways. It will lead to new machine translation quality estimation techniques that take into account the impact of errors on end-users; it will expand the scope of explainable artificial intelligence research to encompass the considerable risks and harms caused by language generation tools, and it will generate new interface design that assists lay users' sense making of artificial intelligence systems.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.