SAFE-AI Task Force Expands Ethical AI Guidance to Include Evaluation Criteria for Automated Interpreting Solutions

The SAFE-AI Task Force has announced the release of a new report that offers organizations a comprehensive framework for evaluating the deployment of automated interpreting solutions. Developed by CSA Research on behalf of SAFE-AI, the report introduces a six-dimension assessment rubric designed to help organizations assess both the benefits and potential risks associated with Automated Speech-to-Speech Interpreting technologies, also referred to as automated interpreting by artificial intelligence (AIxAI).

The new report, Automated Speech-to-Speech Interpreting: Six Evaluation Dimensions for Professional Deployments, serves as an extension of the official SAFE-AI Task Force Guidance on AI and Interpreting Services issued in June 2024. This addition provides organizations with practical tools for evaluating automated solutions, with specific focus areas including input-output reliability, technology capacities, alternative solutions, and the impacts—both intended and unintended—of implementing these technologies.

“The six dimensions in this report are designed to guide organizations in making informed decisions about when AI-driven interpreting can be beneficial and when it poses too many risks to proceed,” said Holly Silvestri, SAFE-AI Task Force spokesperson. “By addressing critical areas like end-user benefits, operational capacities, and ethical considerations, the report serves as an important tool for safeguarding fairness and accountability in AI applications for interpreting.”

While the report acknowledges the potential of AI in professional use cases—ranging from business-to-business and business-to-consumer interactions to government services—it emphasizes a cautionary approach. It encourages organizations to thoroughly evaluate AI interpreting solutions, particularly in live interactions where risks may be heightened. Importantly, the report highlights that organizations are still responsible for complying with legal and regulatory requirements for language access.

The release of this report is part of the Task Force’s ongoing mission to ensure the safe, ethical, and equitable use of AI in interpreting services. The National Council on Interpreting in Health Care recently adapted the Task Force’s guidance in their working paper, Guidance For Healthcare Organizations Evaluating the Potential Use of AI-generated Interpreting. Additionally, the recently rebranded Coalition for Sign Language Equity in Technology, formerly the Advisory Group on AI and Sign Language Interpreting, will host a day-long workshop (in-person and live-streamed) on AI, Interpreting, and Language Access on November 13, 2024, immediately prior to the biennial meeting of the Conference of Interpreter Trainers (CIT).

Publication Details: The report, Automated Speech-to-Speech Interpreting: Six Evaluation Dimensions for Professional Deployments, is available on CSA Research’s platform as a free-access resource. In addition, a 90-minute webinar discussing the report will be hosted by CSA Research on September 19, 2024, at 10:00 AM CDT. Access to the webinar is free, and registration details are available on the SAFE-AI Task Force and CSA Research websites.

For more information on the SAFE-AI Task Force and to access the latest guidance on AI in interpreting, visit www.safeaitf.org.

MultiLingual Staff
MultiLingual creates go-to news and resources for language industry professionals.

RELATED ARTICLES

Weekly Digest

Subscribe to stay updated

 
MultiLingual Media LLC