UAE's Mohamed bin Zayed University Launches Jais 3: AI That Understands Every Arabic Dialect

An innovative Arabic AI tool, designed to comprehend diverse regional dialects and adapt to evolving language, has been introduced by a leading university in Abu Dhabi. Jais 3, developed by the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) in collaboration with Abu Dhabi's Inception42, is capable of handling inquiries and performing tasks such as transcribing conversations and generating documents and presentations. It incorporates Jais Speech, a voice recognition model specifically trained for the Arabic language and its many dialects across the Gulf and North Africa.

This advanced AI platform was created by the university's Institute of Foundation Models (IFM) with the objective of keeping pace with the linguistic intricacies of a language spoken by over 400 million people globally. The technology also addresses code-switching, the phenomenon where bilingual individuals alternate between languages within a single conversation.

"Arabic AI needs to understand the language in its everyday context, which varies by geography and across generations," explained Haonan Li, senior research scientist at IFM. "For example, we need to account for the many nuances in spoken Arabic, including when speakers slip into English or French mid-sentence and back again. We built Jais 3 for Arabic as it's really used."

High-tech training

Jais Speech can transcribe Modern Standard Arabic and dialects from 19 Arab countries. It accurately follows speakers as they switch between Arabic and other languages like English or French, preserving the original script of those words. Furthermore, it identifies the specific dialect being used and can add Arabic vowel marks to transcripts. When integrated with the language models, Jais Speech empowers Jais 3 to execute spoken commands in either Arabic or English.

Mr Li highlighted the team's extensive efforts to ensure robust representation of the various Arabic forms spoken regionally and internationally. “One of the key challenges was ensuring broad and balanced representation across regional dialects,