🔬 Research summary by Connor Wright, our Partnerships Manager.
[Original paper by Nagla Rizk]
Overview: The Middle East and North Africa region could prove an exciting incubator for the positive use of AI, especially given how different countries have prioritized its development. However, the tendency towards economic growth over social service improvement seems to have deprived a generally youthful population of expressing their digital literacy skills. Regional peculiarities complicate this further, requiring a conducive environment to develop AI within the region becoming ever more apparent.
Could the Middle East and North Africa (MENA) region be a promising region for AI to be used for good? Within the MENA region, AI can exacerbate current social and economic divides by concentrating power in the hands of the few but also help smaller businesses gain access to the market. Nevertheless, as a region that has not benefited from trickle-down economic policies, the multifaceted and multilayered inequality, stretching from inequality of opportunity associated with religion and class to economic inequality, has the potential to get only deeper. The improvement or worsening of this situation is tied heavily to the current data situation in the region, with the current attitude towards education and regional peculiarities complicating matters even further.
The first thing to note within this scenario is the particularities of the MENA region itself. Boasting an oasis of different cultures, dialects, and traditions, it also contains a wide-ranging scale of economic prowess and sources of income. Countries such as the UAE can dedicate more resources to the development of AI systems (even having a designated government minister to AI). In contrast, others, such as Mauritania, just don’t have the necessary means to do so.
Nonetheless, what unites the region through a common thread is the youthful population present within each country and the relentless pursuit of tech development and economic growth being the norm. This has meant that aspects such as economic development, inclusionary practices, political participation, and other social service development have not reached the top of any political agenda. One significant cost being education.
The need for education
An overarching acknowledgment in AI is the need for specific education surrounding the area to better prepare the population for adapting to the current times. However, despite many countries within the region’s working population being included in the bracket of medium-low skilled workers (whose jobs are most at risk of automation), projects to up-skill such workers have not been fully pursued. This is certainly not the case because of a lack of digital literacy within the countries involved. Most of the populations of said countries have internet access and are familiar with its usage.
However, such skills cannot be utilized by said populations, with mass unemployment running rampant through the region. One potential solution can be to resort to entrepreneurship to take advantage of these digital skills, but a severe lack of data accessibility amongst other factors (such as non-affordable cloud data structures) proves severely debilitating for efforts to do so.
With the states within the region being primarily in control of the national statistics centers that harbor most of the country’s data, the lack of available open data then inhibits young entrepreneurs in these regions from being able to further their own projects. Given how most data collection is then centralized, extremely accurate local data is harder to overcome through a lack of local data repositories and initiatives. This then brings to the surface how if there is now locally centered data collection, how representative and accurate will the data be?
With a centralized role being played by the state in these regions, there runs the risk of minority and more micro-level experiences being overlooked. For example, data being collected by the state about hard-to-reach rural communities (who do not enjoy a high level of internet access) is unlikely to be carried out to such great lengths to compensate for the lack of access. In this sense, it is likely that the data that is being collected on these areas through the areas that are slightly more connected are then taken as representative of the whole of these regions. Certain inaccuracies about the experience of such communities can then lead to some inaccurate assumptions about what those experiences consist of. These assumptions would then be translated into the AI design itself.
The major problem is how a country-wide AI policy or usage cannot represent all thanks to the presence of these assumptions, being labeled as “data blindness” in the paper. In this sense, the data is ‘blind’ through not fully representing the community it is being utilized to describe, contributing to a “data myopia.” In this sense, the paper notes how national, aggregate-level methodologies for data collection cannot accurately reflect reality when it only reflects the existence of a particular segment of the population.
Between the lines
It is important to note, still, that the MENA region cannot be treated as a singular entity. Various countries such as Tunisia, Jordan, and Egypt are all at different legislative stages of the AI and data process, with the uneven economic development mentioned beforehand playing a part in this. In this way, only some members of the MENA region can take advantage of implementing AI positively. For me, lacking in education or digital literacy is not the most significant problem in allowing all in the region to take advantage of AI, but the lack of opportunity to express such skills in the work environment. More investment is not to be seen as the panacea but rather the construction of an environment conducive to innovation and development that will genuinely allow the region to take off in the AI space.