Paper: AI models exploit translators' work as data without credit

A new paper explores how translators' work has become a foundational data source for AI, particularly in machine translation. The research highlights that translation memories and parallel corpora, while crucial for training AI models, are often acquired without proper attribution or compensation to the translators. The paper introduces concepts like "appropriation without consumption" and the "invisible teacherisation" of translators to describe this process, examining legal frameworks and data supply chains to propose redistributive design solutions. AI

IMPACT Highlights ethical concerns regarding data sourcing for AI, potentially influencing future data collection and compensation models for human labor.

RANK_REASON The cluster contains a single academic paper discussing AI and data ethics. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• AI
• Copyright law
• European legal frameworks
• Japanese Copyright Act
• Large language models
• machine translation
• Translation memories
• United States legal frameworks

• paper
• policy
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →