As artificial intelligence reaches its next major inflection point, it also faces significant challenges to its advancement.
Today’s AI systems can already draft emails, analyse data and hold conversations, but on the horizon lies a more advanced frontier: Artificial Super Intelligence (ASI), a form of intelligence that could surpass human capabilities across virtually all domains.
Before we can assess the implications of ASI, we need to understand the current path we’re on. The vast majority of AI systems available today, including virtual assistants like Siri and Alexa, as well as customer service chatbots, fall under what’s known as Narrow AI.
These systems are designed to perform specific, pre-defined tasks. While impressive, they lack true adaptability, contextual understanding or the ability to generalise knowledge across different domains.
AGI is the next evolution of AI
The next major leap in AI is the emergence of Artificial General Intelligence (AGI). AGI refers to systems capable of performing any intellectual task that a human can, with similar versatility and reasoning ability.
Imagine a CEO working alongside an AGI assistant that not only helps devise a product launch strategy, but also drafts legal documents and offers strategic advice, all within the flow of a single, seamless conversation.
Beyond that, however, lies ASI: an intelligence not just equal to humans, but far exceeding us in logical reasoning abilities, creative thought, problem-solving skills and emotional intelligence.
Such an artificial system would not only execute our assignments, but also enhance our work practices and discover unanticipated problems while developing superior frameworks than any human could conceive.
Experts suggest that AI will evolve from AGI to ASI through a process known as recursive self-improvement – a self-reinforcing loop in which an AI system enhances its own code, accelerating its intelligence, speed and efficiency with each iteration.
The line between a brilliant intern and a visionary CEO begins to blur when you introduce an AI that doesn’t just outperform human leaders but redefines what genius looks like altogether. Yet the path to this future is not equally paved for all.
In the Arab region, a major obstacle remains: the underdevelopment of AI systems in Arabic.
While English and other dominant languages enjoy advanced linguistic modelling, Arabic, spoken by over 400 million people, is still underrepresented in the AI landscape.
This linguistic gap could limit the region’s ability to participate fully in the next wave of AI evolution. Why is this?
Morphological complexity
A single root can generate dozens of distinct word forms, making pattern recognition, translation and natural language processing challenging for AI systems.
Dialects vs standard Arabic
The gap between Modern Standard Arabic and regional dialects is vast. Dialects differ so drastically across regions – for example, Moroccan Arabic and Gulf Arabic are almost mutually unintelligible – that training a single AI model to understand “Arabic” is akin to teaching it several languages at once.
Contextual nuance lost in translation
Literal translations often miss cultural and contextual cues. Arabic is highly context-dependent, and direct translations can fail to preserve meaning, intent, or tone, especially in professional, legal, or creative content.
Thus, an artificial intelligence system needs advanced capabilities to grasp Arabic at levels needed for business deals or diplomatic communication, as well as poetic interpretation.
Today’s existing narrow AI struggles with this kind of work, while AGI would require thorough training to handle it.
The question is no longer whether ASI can speak Arabic like a native – the real challenge is how we will train it to do so
Three essential developments must occur for the Arabic language to enter the era of AGI fully and then progress to ASI.
Comprehensive, structured access to Arabic data
To train models that understand the complexity of Arabic, developers need structured access to a wide corpus of language materials. This includes spoken dialogue, regional dialects, classical Arabic, digital slang and even meme culture. Without this linguistic depth and diversity, AI systems will remain shallow imitators, unable to grasp the full spectrum of Arabic expression.
Robust AI infrastructure
Progress requires infrastructure and investment. Saudi Arabia and the UAE are currently leading the charge, with models like Allam and Falcon reflecting serious national commitments.
Deeper collaboration with agile private sector innovators is essential. A standout example is Pronia, the Arabic LLM developed by Arabic.ai – a promising sign of what private R&D can deliver when given the space and support.
Training AI to grasp cultural context
Arabic is more than syntax and grammar – it’s culture, tone, subtext. To achieve real fluency, AI must learn the cultural cues embedded in everyday communication. Without this, Arabic-speaking AI agents will remain tone-deaf and transactional, far from the relational intelligence that AGI and ASI require.
When ASI masters Arabic, it will create immense value throughout the Gulf region and worldwide.
However, reaching that point demands focused effort, deep cultural understanding, and bold, forward-thinking strategies.
The question is no longer whether ASI can speak Arabic like a native – the real challenge is how we will train it to do so, and how soon we begin.
Yousef Khalili is the global chief transformation officer and CEO MEA at Quant
