A/HRC/57/70 41. There is a strong correlation between poverty and the level, quality, and quantity of education, and correspondingly between racialized poverty and racialized education. The digital divide in education is generally aligned with this existing divide. In poor neighbourhoods and communities, there is less access to digital education from nursery through post-secondary levels because of limited access to electricity or alternative energy sources, to digital devices, to digital content and, ultimately, to teachers and university professors who themselves are digitally competent. Where people of African descent are poorer than other groups, the intersection between poverty and race has resulted in digitally disadvantaged homes, schools and communities. This means that children of African descent enter higher levels of education and training disadvantaged and consequently are less prepared to work in the digital world. 42. At higher levels of education, race plays a triple role in digitalization and the use of AI in two ways. First, there is much less research about the issues and concerns of people of African descent, including and particularly by scholars and researchers of African descent. Consequently, the frameworks used for such research may, intentionally or unintentionally, introduce and perpetuate racial bias and prejudice, rendering the data they produce injurious to people of African descent. Second, because AI is predicated on the use of large data sets; the paucity of accurate data about people of African descent presents another layer of bias. Third, biases and stereotypes are deeply embedded in machine learning, 23 as for example in photography where “photographic systems attempted to create a universal or neural standard for all subjects, yet the norm ended up being white skin”. 24 43. In its 2021 flagship report ‘World Employment and Social Outlook: the role of digital labour platforms in transforming the world of work’, the ILO documents the pervasive nature of digitalization and AI in virtually all aspects of work from education to agriculture, transport to industry, and the ubiquitous nature of all types of learning platforms.25 Concerns about the exposure and loss of jobs because of digitalization and AI are justified, as experts and researchers have exposed in some areas. Routine jobs, depending on the overall development, wealth and technological and industrial advancements of the society, are particularly vulnerable to automation.26 The researchers also indicate that machine learning systems are also able to improve performance in non-routine tasks. They indicate that in lowincome countries, only 0.4 per cent of total employment versus 5.5 per cent in high income countries is likely to be affected by automation. They flag that the impact of augmentation is higher: 10.4 per cent in low-income countries and 13.4 per cent in high-income countries. This impacts people of African descent. 44. The Committee of Experts on the application of Conventions and Recommendations, in their General Observation (2019) on the Discrimination (Employment and Occupation) Convention, 1958 (No. 111), noted that under the Convention, the term ‘race’ includes any discrimination against linguistic communities or minority groups whose identity is based on religious or cultural characteristics or national or ethnic origin, with ‘colour’ being one of the ethnic characteristics. The Committee acknowledged the persisting patterns of discrimination on the grounds of race, colour and national extraction.27 45. Researchers note that the impact of digitalization and AI on employment is not an apocalypse but a shift, arguing that the potential augmentation effects are higher than 23 24 25 26 27 Ludovica Marinucci1, Claudia Mazzuca and Aldo Gangemi, “Exposing implicit biases and stereotypes in human and artificial intelligence: State of the art and challenges with a gender focus”, AI and Society, 38 (2): 747-761 (2023); Ryan Baker and Aaron Hawn, “Algorithmic Bias in Education”, Algorithmic Bias in Education, International Journal of Artificial Intelligence in Education, 32(4), (2021). Nettrice R. Gaskins, “Interrogating AI Bias through Digital Art. Social Science Research Council”, Just Tech, 7 September 2022, available at https://doi.org/10.35650/JT.3039.d.2022. ILO, 2021. Submission by ILO, Paweł Gmyrek, Janine Berg and David Bescond, “Generative AI and jobs: A global analysis of potential effects on job quantity and quality”. ILO Working Paper 96. (Geneva: ILO, 2023). ILO, Committee of Experts on the Application of Conventions and Recommendations, 2019. 11