By Sulamaan Rahim and Gonzalo Grau

In the 2023 edition of the AI Readiness Index, Egypt has emerged as a regional leader among its North African peers. It is the first MENA country to adhere to the OECD Principles on Responsible AI and has adopted a proactive approach in shaping the incorporation of AI into its national economy and public service infrastructure. It aims to use it to bolster its existing strategic assets and boost its capabilities with regards to the Sustainable Development Goals (SDGs). This focus is reflected in its National AI Strategy, which has encoded sustainability into its vision through its four pillars: AI for government, AI for development, capacity building, and international relations. Egypt’s efforts to incorporate AI into its public infrastructure are not limited to strategy, however. Concrete measures for implementation as well as the development of domestic AI applications are also on the agenda.

Dr. Hossam Osman, advisor to the Minister for Technology Innovation, Electronics Industry, and Training at the Egyptian Ministry of Communications and Information Technology (MCIT), spoke to us about Egypt’s priorities, projects and challenges in relation to the strategy.

AI for Human Development: Egypt’s capacity building approach

Egypt’s National AI strategy describes human capacity building (AI4H) as its most important and most difficult-to-implement pillar. Dr. Osman suggests that human capital is the country’s biggest real strategic asset. Egypt is, and always has been, about its human assets; it is already one of the most attractive economies for outsourcing, placing 3rd in the 2023 Global Outsourcing Confidence Index. 

Its approach to capacity building for AI adopts a sociotechnical understanding of AI, acknowledging how commonplace systems will be in the near future. Adopting a pyramid-shaped set of initiatives, the strategy targets both technical and non-technical roles within the AI ecosystem. Another pyramid also illustrates the important actors within the ecosystem; increasing in specialisation as you go up the pyramid, this demonstrates the fundamental importance of AI understanding within the general public  as well as the need for more generalised roles such as software developers to ensure AI capabilities.  

Source: National Council for Artificial Intelligence 

The initiatives taken at each level of the pyramid are tailored to a group’s specific needs, skills, and level of understanding with respect to artificial intelligence. The lower levels focus on awareness. Here, work is being done to enhance AI awareness within the public sector: over 5000 government officials have been trained to understand and explain use cases for AI. Structurally, the Egyptian government has introduced a digital transformation unit within each of its ministries, all of which are staffed with professionals fully trained in AI use cases. 

Moving up the pyramid, the initiatives take on a more technical colour and are tailored to higher skilled groups. The Digital Egypt Cubs Initiative (DECI), launched in May 2022, aimed to equip school students with fundamental IT skills, including AI. The initiative targets 3,000 top performing students at the national level, and is done in collaboration with international entities that specialise in AI related skills. Egypt Future Work is Digital (FWD) is a fully funded digital upskilling scholarship program for four skill levels (courses are tailored from the foundational level to the expert level). The program provides a range of AI-related specialisations including cybersecurity, embedded systems, cloud computing, and more. The program, according to Dr. Osman, aims to upskill students and professionals in AI and related technologies so as to better match market needs at both the international and domestic level, though Egypt’s already strong capacity in freelancing and outsourcing mean that international demand is prioritised. Efforts are also being made to ensure that special attention is paid to AI in higher education, with the MCIT establishing the specialised Egypt University of Informatics (EUI) in 2021.

At the top of the pyramid sit initiatives like the Digital Egypt Builders Initiative (DEBI), a scholarship launched by the Ministry of Communications and Information Technology. The initiative targets leaders in the ICT field, offering a professional master’s in AI and emerging technologies at top international universities, as well as professional training by local and international tech firms. These initiatives are just a few examples among many, which reveal Egypt’s strong commitment to a human-centred AI strategy. 

Challenges to implementation: public perception, data, and compute

A focus on sustainable AI will of course encounter some challenges and AI strategies in general can be difficult to implement. Dr. Osman identified public perception as one of the main challenges Egypt faces when it comes to AI uptake. Widespread fears around automation within government are already being addressed, but easing these worries in the private sector is proving to be more challenging. Despite this being accounted for in Egypt’s capacity building pillar, where awareness is of pivotal importance, shifting public opinion towards one that is ready to embrace AI will be a gradual, difficult process. 

Another major challenge relates to data. Being the fuel behind all AI systems, ensuring the numerous security issues associated with data are dealt with is paramount. Anticipating the need for the right data capabilities, Egypt has—pending the establishment of a permanent governing body—created an affiliate ministerial entity responsible for all data-related policy issues. In particular, the entity will assess the level of data maturity across the country’s industries and organisations.

In terms of legislation, there is some room for improvement with regards to AI. Despite having a data protection law in place since 2020 and having completed drafting a data sharing law, Egypt does not yet have the accompanying executive legislation needed for it to be implemented rigorously. Dr Osman points out that this is a general issue in national AI policy — some general legislative structure or law may be required to guide and provide strong foundations for the implementation of the strategy.

Finally, Dr. Osman identified compute infrastructure as the last major obstacle to a smooth implementation of Egypt’s AI strategy. When talking about AI adoption within public infrastructure, it is vital to have a domestic level of compute infrastructure to support it. Generating a wide array of applications entails a larger amount of data to be processed, and this must be done domestically to minimise risk and security threats. Egypt is currently planning a series of large investments in expanding its domestic compute stock.

Leading the way: the Egyptian Charter for Responsible AI

After being the first MENA country to adhere to the OECD Principles on Responsible AI, Egypt has just released its Responsible AI Charter.  Implementing the charter falls neatly into the National AI Strategy. As part of its second phase, the government has begun working to implement it with domestic AI developers and practitioners. Through their national database of domestic IT companies, the Information Technology Industry Development Agency (ITIDA) has begun diffusing the charter’s requirements through the country’s ICT industrial network. For companies developing AI products, ITIDA will assess their testing and development teams against the charter’s key principles. Additionally, a self-assessment toolkit is being developed for open delivery to AI developers. The toolkit aims to facilitate responsible AI assurance for small and medium operations, particularly startups and scaleups that may struggle to take on otherwise costly compliance requirements. 

Dr. Osman notes the influence of other guidelines in the charter’s design, particularly the OECD and UNESCO guiding principles, as well as the EU’s AI Act and the US AI Bill of Rights. It includes annual revisions to account for new developments in AI. Though implementation of the charter has begun, it is still in early stages: current efforts aim to fully integrate assurance into the service delivery process.

‘Mature’ AI applications and Egypt’s plans for developing a domestic LLM

Dr Osman reiterated that Egypt’s ambitions with regards to AI also pass through the development of ‘mature’ AI applications. By this, Dr. Osman is referring to an application that is based on the availability of local data that is representative of the Egyptian population. Currently, Egypt lacks the level of data openness required for the free circulation of this data between institutions, he says. A mature data scheme involving an open data platform would enable institutions to use and exchange local data for the training of applications that reflect the country’s context. The potential added value of localised AI applications that are fed through open data is enormous. When a local data approach was adopted in training an application for the detection of diabetic retinopathy (which, untreated, can cause blindness), it was found that the performance was much better in Egypt when compared to other countries. 

Localisation also extends to the newest development in AI, large language models (LLMs). As part of its vision of harnessing AI for boosting capabilities for outsourcing, Egypt has begun to define its approach for integrating LLMs into its national AI infrastructure. Indeed, technology — in addition to data and compute infrastructure — is a key part of the fundamentals driving Egypt’s implementation strategy, and LLMs can be leveraged just as much as other AI applications to help boost capabilities in its key sectors. As such, one of the strategy’s most innovative dimensions proposes the use of large language models for top priority industries in Egypt. The debate therefore revolves around the way this is to be done. The options are threefold, and come with different degrees of difficulty as well as scale. If they are found to be adequate, Egypt could opt for fine-tuning the parameters of existing models without altering their structure. Should this not be sufficient, structure alterations to the model are also within scope. Alternatively, Dr. Osman points to the development of an entirely new, domestic large language model. 

Looking ahead

Egypt has shown initiative and creativity in building its AI strategy. The strategy is designed to play into the country’s strengths (offshoring, outsourcing), all the while paying attention to areas for improvement, such as public perception and understanding. Taking a proactive approach to AI uptake ensures that key aspects of its sustainable use are focused on, such as data openness and localised applications. 

Recently, Egypt has begun discussions to implement the second phase of its strategy. The first phase was finalised in May 2024, and left room for a number of measures across six different dimensions over the next three years: governance, ecosystem, compute infrastructure, data, human resources, and technology.

Cover photo by Yousef Salhamoud on Unsplash