From Code To Conversations

A look at the advances Saudi Arabia is making in the field of prompt engineering, which is essential for the democratization of artificial intelligence

Artificial intelligence (AI) has transformed by leaps and bounds since the Turing Test in 1950 and since the term was first used in 1956. At the same time, the ways in which humans interact with such AI systems have evolved too. 

Among the most recent developments in these interactions is prompt engineering, which refers to the skill by which AI models are guided and utilized. This is becoming increasingly important as AI tools like ChatGPT, DALL·E, and others gain widespread adoption due to the role of prompts in unlocking the full potential of these technologies. These platforms require precise and well-worded instructions and questions to produce optimal results. 

Therefore, in Saudi Arabia, prompt engineering is set to feature as a critical tool in the country’s technological ambitions, especially as technology and innovation are being prioritized by the country’s Vision 2030. 

Understanding Prompt Engineering 

Prompt engineering refers to the way in which people interact with AI, with the practice moving away from traditional programming to natural language-based communication. Unlike conventional software programming, which often requires specialized skills in coding languages, prompt engineering allows non-technical users to use AI as well. 

“Instead of writing code in specialized languages, you can now use everyday language – like English or Arabic – to instruct AI models,” explained Hassan Sawaf, founder and CEO of San Jose-headquartered aiXplain, Inc. “This shift began with tools like ChatGPT and continues with agentic AI platforms such as aiXplain. Imagine being able to ‘program’ entire business units simply by conversing in your preferred language… This is no longer a distant vision. You can now integrate these capabilities into your organization by learning how to efficiently use prompt engineering to empower your workforce.”

This shift is particularly of importance in sectors where innovation is critical, such as healthcare, education, e-commerce and finance. For instance, in education, prompt engineering can enable AI to create personalized curriculums for students based on their individual strengths and challenges. In healthcare, it can assist with streamlining diagnostic processes, and enhance patient care through AI-driven insights. 

“Prompt engineering makes programming accessible to everyone in a company, removing bottlenecks and speeding up research and development,” Sawaf explained. “Instead of waiting months for specialized engineers, domain experts – whether they’re in healthcare, e-commerce, or education – can directly craft AI solutions themselves.” 

Why Prompts Matter 

Saudi Arabia’s Vision 2030 places significant emphasis on technology as a driver of economic diversification. The Kingdom’s investments in AI infrastructure, research, and education are a sign of its commitment to becoming a global leader in this space. Prompt engineering, with its ability to lower barriers to AI adoption, makes these goals all the more achievable. 

Bernard Ghanem, Chair of the Center of Excellence (CoE) for Generative AI (Gen AI) at King Abdullah University of Science and Technology (KAUST), highlighted that his entity was launched to unlock the “significant potential of Gen AI.” The next generation Gen AI models from this Center will be launched in a manner that will allow their widespread use, while also being accessible and trustworthy. 

Ghanem pointed out that the CoE plans to develop GenAI models that overcome the current technical challenges hindering current models in the market. “These models will be customized to drive real-world impact in specific domains of interest, namely, to solve difficult problems outlined in the four research, development, and innovation (RDI) pillars of the Kingdom. For example, our Gen AI models will be used to develop and deploy intelligent solutions for the healthcare sector, where both healthcare providers and the general public can benefit from Gen AIenabled systems.” 

Ghanem continued, “Gen AI is also set to revolutionize the energy sector by providing automated solutions that can help discover new materials with desirable properties that are inaccessible with current labor-intensive and manual discovery pipelines. Regarding sustainability, Gen AI can be harnessed to help automate earth observation, agricultural assessment, and coral reef monitoring. Furthermore, our Gen AI models will drive impact in the economy of the future, with emphasis on creative content generation (e.g. generating and editing images/videos/3D content with minimal human intervention) and education (e.g. real-time AI tutor).” 

Ghanem also revealed that the wider adoption of Gen AI technology in Saudi Arabia will also be championed through the development of a userfriendly platform, where non-experts can quickly deploy and use Gen AI models, while safely using their own data. 

Language And Cultural Nuances 

It’s important to note that while prompt engineering holds immense promise, it also presents a singular challenge, particularly in a nation like Saudi Arabia, where Arabic is the dominant language. This is because AI models are often optimized for English, and Arabic – especially with its varied dialects – can pose a hurdle. 

But there are solutions in the offing. “The good news is that Arabic language capabilities for these models and platforms are rapidly improving, supported by investments in Arabic large language models (LLMs) and foundation models,” said Sawaf, whose platform, aiXplain, has been built to cater to the nuances of the Arabic language. “Locally funded initiatives in Saudi Arabia, for instance, focus on ensuring better cultural and linguistic alignment.” 

Similarly, KAUST’s CoE for Gen AI has made Arabic LLMs an important focus area. In fact, the CoE is building development programs needed to understand, analyze and extract insights from the entire pipeline of building an LLM for the Arabic language. However, KAUST’s Ghanem highlighted the challenges facing this initiative, with one being the limitation of native Arabic data available online and from other sources. 

“This data limitation is even a bigger issue when expanding beyond MSA (modern standard Arabic) and dealing with the many Arabic dialects prevalent in the region and the Arab world,” Ghanem explained. “This challenge can be addressed by standardizing and combining Arabic data crawling efforts, as well as, leveraging non-conventional and innovative means of crowdsourcing this data from users (e.g. through traditional/social media campaigns).” 

Ghanem pointed out that another hurdle is related to specializing building blocks in the LLM pipeline to the Arabic language, including a specialized tokenizer. “This specialization requires retraining or trainingfrom- scratch of LLM architectures, which in turn requires a high level of development skills for scalable operations,” he said. “KAUST’s GenAI CoE is building training programs to make such skills available to developers in the Kingdom.” 

There is also another challenge that lies in prompt design that is aligned with the Arabic culture and adheres to cultural norms. “While translating prompts from other languages may produce reasonable results in some Arabic LLMs, it does not necessarily rise to the level of AI alignment required of a general purpose LLM that is respectful of its cultural and geographical context,” Ghanem explained. “This provides an opportunity for researchers to develop methods and evaluation criteria for prompts that produce aligned Arabic LLM results.” 

As AI models become more sophisticated, the field of prompt engineering can thus be expected to continue to evolve. “However, research over the last couple of years has innovated some techniques that can automate most – if not all – of this prompt design process,” KAUST’s Ghanem pointed out. “In fact, prompt learning is a growing field of research in AI, whereby AI models are trained to ‘learn’ how to prompt LLM models. Additionally, multi-agent systems, where multiple AI models work together to solve a query, are gaining traction. “Some of these innovations – like CAMEL and GPT Swarm – originated from KAUST,” Ghanem noted. 

Speaking about the skills needed to work in the line of prompt engineering, Sawaf said that using natural language to direct agentic AI systems is essential. “The most effective prompt engineers excel at clear, concise communication – so, skills in management, human resources, communication, and problem-solving are incredibly valuable. Think of working with AI as recruiting a team member: you provide the role description (the prompt), offer examples and strategies, then give constructive feedback to improve its outputs. There are training programs designed to help you develop and refine these skills, in particular learning from real-world use cases, so you can become a better manager, trainer, and educator of AI through prompt engineering.” 

Ghanem said that currently, there is still need for intelligently designed prompts that have the ability to extract the precise information needed to answer a particular query. “These prompts have to also be designed in such a way that AI safety guidelines are respected,” he explained. “Manual interaction with particular LLMs is still key to the design of impactful prompts. This is why the role of prompting engineers remains a well sought-after position in companies seeking to deploy Gen AI tools within their operations.” 

Looking to the future, Sawaf noted that the democratization of AI through prompt engineering will empower individuals to become innovators in their own right. Learning this skill will therefore put the power of AI in the hands of the many – rather than just of a few. “Our region is brimming with entrepreneurs and professionals who have amazing ideas,” Sawaf said. “Historically, turning those ideas into technological solutions was challenging. Now, with agentic AI and prompt engineering, anyone – entrepreneurs, artists, lawyers – can serve as their own ‘engineers on the side.’”

Building Blocks 

South Korea-based AI chipmaker Rebellions on its foray into the KSA 

While a number of initiatives and platforms need to be in place for AI solutions, infrastructure is also critical to their success. This remains the key focus, then, for companies like Rebellions – this South Korea-based AI chipmaker has partnered with the likes of Aramco and Wa’ed Ventures to build scalable, energyefficient AI systems tailored to the region’s needs. 

Sunghyun Park, co-founder and CEO of Rebellions, said, “Rebellions is committed to contributing to Saudi Arabia’s Vision 2030 by driving innovation in AI infrastructure and enabling technological advancements across key industries. Our approach focuses on developing high-performance AI chips and systems that support large-scale AI deployments, fostering digital transformation throughout the region.” 

This infrastructure, Park noted, had to be created to address the region’s specific requirements. “As a first step, we signed a memorandum of understanding (MoU) with Aramco in September 2024, focusing on potential AI chip deployment in Aramco’s data centers, and we are currently collaborating on initial steps,” he said. “By leveraging our expertise in AI and partnering with entities like Wa’ed Ventures, we seek to empower businesses, enhance public services, and build a robust foundation for Saudi Arabia’s AI ecosystem, ultimately contributing to the country’s long-term growth and technological leadership. The partnership with Wa’ed Ventures and the establishment of a local branch significantly influence our approach to AI development in the region. By collaborating with Wa’ed Ventures, we gain access to valuable local market insights, networks, and resources that enable us to tailor our solutions to Saudi Arabia’s unique needs and Vision 2030 objectives.” 

Park also highlighted the importance of talent development in the Kingdom. “Establishing a local branch enhances our ability to engage directly with stakeholders, understand their challenges and provide customized AI solutions that align with the region’s priorities, such as digital transformation and AI infrastructure development,” he said. “Rebellions’ local branch hopes to collaborate with educational and research institutions to cultivate local talent, enabling joint development of AI infrastructure within Saudi Arabia’s technology ecosystem. To meet market demands, we plan to invest in local talent development, foster partnerships with key organizations, and focus on creating scalable, high-performance AI solutions that address the specific requirements of industries in Saudi Arabia.”