Artificial Intelligence (AI) and the Evolution of Digital Divides

Gaps, divides, and splits are a common feature of contemporary societies and economies. Both at the global and the national level, inequalities persist across multiple social dimensions, interacting and intersecting in complex ways. From income to education, to digital gaps, divides shape an uneven playing field where access to resources and opportunities are not evenly distributed. With the rapid advance of information and communication technologies, inequalities have continued to reproduce and, in many cases, amplify. Despite efforts to close digital divides, their contours keep evolving in parallel with the rapid technological transformation

The development and deployment of artificial intelligence (AI) systems — including technologies such as machine vision, machine learning, natural language processing, and sophisticated algorithms — across different social domains, it is likely to deepen some of the existing inequalities at the global, regional, and national contexts. Hence, it is crucial that governments, universities, international organizations, and other institutions consider strategies and policies to shape the impact of AI, and co-lead a fairer and more inclusive technological transformation.

 

Mind the Gap

The term “digital divide” entered the public discourse in the 1990s as a way of describing disparities in access to Internet connectivity and computer power within the U.S. population. Several years after the High Performance Computing Act — a U.S. bill that aimed to boost Internet infrastructure and scale up connectivity — passed in 1991, the “digital divide” term started to appear in news outlets and politician speeches. As millions of people, particularly those with better socioeconomic status living in urban areas, started to have Internet connectivity at their homes, schools, and libraries, differences in access to computer and Internet were highly noticeable between regions and population groups.

The “divide,” at this moment of time, was used for describing the unevenness of technology access — the split between the “haves” and “have-nots.” Articles in the New York Times such as “A Nation Ponders Its Growing Digital Divide” and “A New Gulf in American Education, the Digital Divide,” for instance, raised concerns about the disparities in access to computers and Internet among schools. Meanwhile, the National Telecommunications and Information Administration (NTIA) published a report in July 1995 called “Falling Through the Net: a Survey of the “Have Nots” in Rural and Urban America,” that identified the “information disadvantaged” as the ones without access to computers and the Internet, mainly the poor and minorities in central cities and rural areas.

Along with a narrative of empowerment and development that highlighted “access to knowledge,” and “entering the information highway,” in the 2000s, bridging the “digital divide” became a global imperative. The Declaration of Principles of the World Summit on the Information Society (2003), for instance, set up bridging the “digital divide” as a developmental goal for all nations. Governments around the world designed policies for connecting people to the Internet and building digital infrastructure that would allow individuals to join the “information society” and avoid being “left behind.” Benefits of this “information society” included the possibility of accessing knowledge and education, economic growth, foster democracy, and participation.

The focus on technology access and infrastructure made the “digital divide” easy to understand, measure, and quantify. It helped governments, policy makers, and researchers understand the diffusion of Internet across entire populations with the idea of universal access that was used for previous technologies, such as the telephone. However, such approach made some stakeholders, particularly policy makers, overlook other dimensions of society that were also important for the adoption and use of technology such as social stratification, and education.

Researchers from different disciplines started to study how disparities in Internet connectivity and computer ownership created a split both at the global scale (e.g., Norris, 2001) and within nations. In North America, Europe, and Australia, for instance, social scientists found that disparities in access were correlated with age, race/ethnicity, gender, education, and income (e.g., DiMaggio et al. 2004 in the U.S; De Haan, 2003 in the Netherlands;  and McLaren and Zappala, 2002 in Australia).

 

In Latin America, researchers focused their efforts in measuring Internet penetration (CEPAL, 2003),  discussing the meaning of the “knowledge society” (Sociedade da Informação no Brasil Livro Verde in Brazil, 2000), and analyzing the social and cultural uses of ICT (Funredes in the Caribbean, 1995; León, Burch, & Tamayo, 2001 in Argentina; Hopenhayn, 2003 in Chile). However, as Trejo (2004) pointed out, research on the digital divide in the Latin American region was dispersed and lacked a common methodology. According to him, there was a “divide” in terms of the study and theorization of the Internet that Latin American researchers didn’t bridge.

From one Digital Divide to Multiple Ones

As access to Internet and computers spread throughout entire populations within a country and regions, it became clearer that the “digital divide” was not only related to technology access.  Researchers transitioned from analyzing who has access to the technology, to investigating other dimensions of access such as skills and competences, motivations, usage patterns, and tried to understand how the Internet was being used in more advantageous or disadvantageous ways by different populations. Moreover, in an effort to update the terminology around the “digital divide,” scholars proposed alternative concepts, such as “digital inequality” (DiMaggio et al., 2004), “digital/technology for social inclusion” (e.g. Warschauer, 2002; Livingstone & Helsper, 2007), and “participation gap” (Jenkins et al. 2006). At the core of such development was an interdisciplinary effort to understand how the rapid diffusion of digital technologies across national and global contexts was reproducing and amplifying existing social inequalities, and how societies could foster the inclusion of all populations groups as they embraced the digital transformation.

Theorization of the “digital divide,” therefore, has become more nuanced and complex. A number of  studies have found that there are gradients not only in the quality of access to technology, but also in the sociocultural practices and skills that people develop (Hargittai, 2011; Jenkins et al., 2006; Van Dijk, 2005; Watkins, 2012), the information they consume and produce (Robinson, 2009; Schradie, 2011), and the outcomes of their usage (van Deursen & van Dijk, 2013; van Deursen & Helsper, 2015).

Instead of one digital divide, researchers have identified multiple ones. Although the divide in access to material technology is still important (referred to as the “first-level divide”), other digital gaps are now widely recognized. The “second-level digital divide,” for instance, indicates the gap in terms of online skills and practices (e.g., Hargittai, 2002; Jenkins et al. 2006). More recently, scholars have also discussed the existence of a “third-level divide” that consists in the differential tangible outcomes that come from technology usage. For instance, the differences in people’s use of the Internet for improving their socioeconomic status, and earning different forms of capital (e.g., van Deursen & Helsper, 2015).

AI: A New Digital Divide?

One of the latest developments in the evolution of information and communication technologies is the deployment of artificial intelligence (AI) at scale. As AI technologies have started to be embedded in culture (e.g. Facebook feed algorithm), education (e.g. automated tutors in MOOCs), health (e.g. treatment of cancer), the economy (e.g. self-driving cars), and other social domains, its development affects the three digital divides that researchers have identified thus far. Given the reliance of AI in big data and computer power, its  diffusion is more likely to amplify existing inequalities, increasing disparities in access to technology, skills, and tangible outcomes of usage. Notably, AI might accelerate the deepening of second-level and third-level digital divides.

Some commentators have pointed out that the diffusion of AI can create a new digital divide (e.g., Dans, 2016). This perspective assumes that access to AI technologies — machine vision, machine learning, natural language processing, and sophisticated algorithms  —  offers an advantage to the companies and firms that are able to deploy them. The AI “haves” can optimize their operations, automate processes, and innovate, while the “have-nots” are being left behind in world’s digital evolution.

The notion of AI as a new digital divide, however, is problematic, since it forces us to focus on the latest technological advances while ignoring their relationship with other digital gaps previously theorized. Moreover, this notion of a new divide focuses only on the disparities among companies and firms, ignoring the role of governments, population groups, and users. They are also important actors in the diffusion of AI that would be affected by automated decision-making processes. They would play important roles generating and curating data for feeding AI systems, and could shape technological transformation with policies, practices, and institutions.

Global Disparities

So far, AI technologies are being developed and deployed faster in the global north, enhancing inequalities among regions. As a recent PwC report shared by the World Economic Forum showed, the regions that would gain the most from AI are China, and the United States. According to the report, AI could boost China’s productivity by 27% by 2035, particularly in sectors such as healthcare, retail, and ICT.  Regions in the global south like Latin America, in contrast, would only increase 5.4% of GDP.

 

Countries who have not bridged the technology, skills, and usage gaps would likely confront the expansion of those splits as AI diffuses around the world. Particularly, it is highly possible that the second and third level digital divides would deepen as the skills necessary to deploy AI technologies are of higher level than the ones needed for previous ICT. Countries without capacity to develop AI technologies would likely have to deploy AI systems developed abroad that perhaps do not adapt that well to their cultural and social contexts, making more more elusive the tangible outcomes of their use. These nations would be in fragile positions to negotiate issues of data ownership with transnational corporations, and would have to fight for a more culturally sensitive building of databases, and algorithms that take into account the characteristics of their populations and geographies.

Conclusion

As AI reshapes the contours of the existing digital divides at the national, regional, and global scales, we should think critically about how these technologies operate across the multiple dimensions of access (technology, skills, and usage). We should also put attention to the transformations that cultural, human, and social systems need to have in order to successfully  and fairly deploy AI technologies. Moreover, it is necessary to discuss how AI technologies would be regulated locally and globally, within a framework of human values and rights that not only respects the diversity of cultures, genders, and races/ethnicities, but also fosters their inclusion.

References

CEPAL (Comisión Económica para América Latina) (2003) Los caminos hacia una sociedad de la información en América Latina y el Caribe, Conferencia Ministerial Regional Preparatoria para la Cumbre Mundial sobre la Sociedad de la Información: Bávaro, Punta Cana, República Dominicana.

De Haan, J. (2003) IT and social inequality in the Netherlands, IT & Society 1(4) (2003), 27-45.

DiMaggio, P., Hargittai, E., Celeste, C. & Shafer, S. (2004). Digital Inequality: From Unequal Access to Differentiated Use. In Social Inequality. Edited by Kathryn Neckerman. New York: Russell Sage Foundation. 355-400.

Hargittai, E. (2011). Digital na(t)ives? Variation in Internet skills and uses among members of the ‘Net Generation.’ Sociological Inquiry, 80(1).

Hopenhayn, M. (2003) «Conjeturas sobre cultura virtual. Una perspectiva general y algunas consideraciones desde América Latina», en Calderon, F. (coord.): ¿Es sostenible la globalización en América Latina? Debates con Manuel Castells, Vol. II, PNUD Bolivia / FCE, Santiago de Chile.

Jenkins, H., Clinton, K., Puruhotma, Robison, A.R., & Weigel, M. (2006). Confronting the challenges of participatory culture: Media education for the 21st century. Cambridge: MIT Press.

León, O., Burch, S., & Tamayo, E. (2001) Movimientos sociales en la Red, Agencia Latinoamericana de Información: Quito.

Livingstone, S. and Helsper, E. (2007). Gradations in digital inclusion: children, young people and the digital divide. New Media & Society, 9 (4). pp. 671-696.

McConnaughey et al. (1995) Falling through the net: A survey of the ‘have nots’ in rural and urban America. Washington, D.C.: U.S. Department of Commerce. National Telecommunications.  Accessed September 1, 2017. https://www.ntia.doc.gov/ntiahome/fallingthru.html.

McLaren, Jennifer and Gianni Zappalà. 2002. “The ‘Digital Divide’ Among Financially

Disadvantaged Families in Australia.” First Monday 7, 11. http://firstmonday.org/issues/- issue7_11/mclaren/index.html. Accessed Sept. 1, 2003.

Norris, Pippa. 2001. Digital Divide? Civic Engagement, Information Poverty and the Internet in Democratic Societies. NY: Cambridge Univ. Press.

Robinson, L. (2009). A taste for the necessary. A Bourdieuian approach to digital inequality. Information, Communication and Society 12 (4), 488.

Schradie, J. (2011). The Digital Production Gap: the Digital Divide and Web 2.0 Collide. Poetics. 39 (2).

Trejo, R. (2004)”La investigación latinoamericana sobre Internet”. Telos, num 61, octubre-diciembre 2004.

van Dijk, J. A. G. M. (2005). The deepening divide: Inequality in the information society. Thousand Oaks, CA: SAGE.

van Dijk, J. A. G. M. (2012) The Evolution of the Digital Divide: The Digital Divide turns to Inequality of Skills and Usage. In J. Bus et al. (Eds.) Digital Enlightenment Yearbook 2012. IOS Press, 2012

Watkins, S. C. (2009). The young and the digital: what the migration to social-network sites, games, and anytime, anywhere media means for our future. Boston: Beacon Press.

Warschauer, M. (2002) Reconceptualizing the Digital Divide. First Monday, [S.l.], july 2002.http://firstmonday.org/ojs/index.php/fm/article/view/967 Acceded on Sept 02. 2017.