Tag: Bias
Towards Inclusive Natural Language Processing
This article explores the linguistic theories that laid the foundation for natural language processing, confronts the ethical risks of allowing machines to speak on our behalf, and presents a new vision for socially responsible AI language models.
Ensuring Quality and Ethics in AI Localization Workflows
As machine translation becomes commonplace, localization teams must ensure that output meets expectations and that processes uphold transparency, accountability, and respect for user privacy.
Instagram Egregiously Mistranslates Palestinian User Bios, Inserting Word “Terrorist”
A bug in Instagram’s auto-translate feature is highlighting the problem of bias in multilingual language models (MLMs) and the need for more transparency from companies that employ them.