The UAE government has adopted a revolutionary artificial intelligence large language model – Jais AI Model created exclusively for Arabic and developed in Abu Dhabi. The goal of creating an AI language LLM was to incorporate one of the world’s most widely spoken languages into the mainstream AI ecosystem.
Meet Jais AI Model, the UAE’s very own LLM
Jais, named after the UAE’s tallest mountain, will bring the benefits of generative AI to the Arabic-speaking globe. This open-source bilingual Arabic-English model, known as Jais, was developed by Inception, a subsidiary of Abu Dhabi’s AI business G42, Mohammed bin Zayed University of Artificial Intelligence (MBZUAI), and Silicon Valley’s Cerebras Systems.
The developers claim that Jais outperforms current Language Models (LLMs) for Arabic. This resource is available for download from the machine-learning platform Hugging Face.
According to Andrew Jackson, CEO of Inception, the launch of Jais acts as a positive step toward pushing the scientific and computational community to devote more attention to non-English LLMs, similar to initiatives witnessed in Japan and India.
In an interview with a local news agency, Jackson elaborated, “We envision Jais as highly valuable for generative applications, such as formulating responses to queries, generating documents, performing translations, composing emails, and even dispensing advice and recommendations.”
According to the partnering firms, Jais can skillfully capture the distinctions inside diverse Arabic dialects. Furthermore, it has the ability to interpret language, context, and cultural allusions. These capabilities make it significantly more exact and contextually relevant than other models.
This innovation, dubbed ‘Jais’ after Ras Al Khaimah, UAE’s tallest peak, was developed solely for government usage, as well as industries spanning finance, energy, climate change, and health services.
: From Silicon Valley to UAE: The journey of ‘Jais’ language model (WION)
Developed mostly for the UAE Government
The Ministry of Foreign Affairs, the Ministry of Industry and Technological Development, the Department of Health – Abu Dhabi, ADNOC, Etihad Airways, FAB, and e&, the technology conglomerate formerly known as Etisalat, have all joined as launch partners for Jais.
Jais underwent training on the Condor Galaxy, the “world’s largest AI supercomputer”. Condor Galaxy’s inception took place in July by G42 and Cerebras. Furthermore, around116 billion Arabic tokens and 279 billion English tokens used in this training. The model is constantly evolving as additional Arabic information is collected to build new instruction sets.
The importance of local languages in LLMs and AI
According to WorldData, over 400 million people speaks Arabic, making it one of the widely spoken language worldwide. Arabic acts as the official language of 22 countries and 11 more countries speaks in it. However, according to data collected from the participating companies, its online presence remains restricted. Sincem only about 1% of Arabic content accessible on the internet.
Jackson stated that Jais would help raise this statistic. Furthermore, he noted, “We’re initiating an project to collect more Arabic data from offline sources. This initiative has already been launched earnestly.”
He added, “We’re also exploring novel methods to synthesize Arabic content and translate existing English content into Arabic. Although we have a long way to go, optimism is crucial as we vigorously advance.”
Overall, this has created a new battleground in the tech industry. Corporations vying for an early advantage and broadening their boundaries in generative AI.
You can download Jais at Hugging Face. Furthermore, you can also register your interest at the Jais website and upon receiving an invitation to the playground environment you can try Jais online . Lastly, you can read the Jais white paper to learn more about Jais and how it compares to other models.