Jais: UAE’s Arabic Language AI Model
Jais is not a French word, though its first impression may seem very French. In reality, it’s an Arabic open-source AI model launched by UAE‘s G42.
This new language model was created with the help of supercomputers produced by the Silicon Valley-based Cerebras Systems.
Why Jais?
It’s named “Jais” after the highest peak in the United Arab Emirates (UAE), Jebel Jais, located in the northern Emirate of Ras Al Khaimah. It’s a 13-billion-parameter model that was fed by a substantial dataset of both Arabic and English texts, due to the lack of Arabic data. The computer codes in the English language helped train a model of such size.
Becoming the largest Arabic language AI model to date, it offers a broad range of missions. These include:
- Translation
- Answering questions
- Generation of texts, creative content, and codes.
As an extreme admirer of the Arabic language, I can say proudly and without blinking an eye that it’s a highly inflectional language. This means that the same word can have many different forms based on the context of the sentence. Additionally, it has a large vocabulary that can’t be translated or found in other languages.
For example, there is no English term equivalent to the Arabic word ‘gaher’. ‘The dictionary says “anger,” but that’s not quite it. It’s when you take anger, place it on a low fire, add injustice, oppression, racism, dehumanization to it, and leave it to cook slowly for a century. And then you try to express it, but no one hears you. So, it sits in your heart, settles in your cells, becomes your genetic imprint, and then moves through generations. And one day, you find yourself unable to breathe. It washes over you and demands to break out. You weep, and the cycle repeats.’
Innovational Wave in Arabic Language
For all Arabic speakers, I’m talking to 400 million people across the world. UAE has made the G42 AI language tool Jais an open-source project, allowing anyone to download and use it to foster their own AI applications.
Andrew Jackson, CEO of Artificial Intelligence at Core42, said, “The launch of Jais 30B is a giant leap forward for the Arabic-speaking world in harnessing the potential of generative artificial intelligence.”
For an Arabic model, there aren’t many solid, resourceful platforms for the 400 million people to use. What UAE did was to focus on developing new Arabic-language AI applications, like chatbots and virtual assistants. This could enhance the contribution to the progression of new AI technologies in the Arabic language.
Why UAE in Particular?
UAE is at the vanguard of Arabic language AI research and expansion. It is well-equipped in AI research. Going back to 2017, the UAE founded the Ministry of Artificial Intelligence, the first ministry of its kind in the world. There is substantial government support to rely on.
Not to forget, UAE has partnered with Google AI to develop the Falcon family of Arabic language AI models. It invests strongly in robust partnerships, giving it an open door to collect data since it has a significant amount of data on the Arabic language, which is key to training AI models.
All data comes from trustworthy sources like government agencies and businesses.
Although the G42 AI language tool is still under development, the model is offering longer and more detailed answers, with an increase of 160% in Arabic and 233% in English compared to previous models.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Tech sections to stay informed and up-to-date with our daily articles.