Sociology & Political ScienceSystematic Review
The Digital Divide in 2025: Why Internet Access Is a Social Inequality Problem, Not a Technology Problem
The digital divide was supposed to close as technology became cheaper. Instead, it has evolved: from a gap in access to a gap in skills, from a gap in skills to a gap in AI readiness. Five papers argue that the divide is fundamentally a social inequality problem that technology alone cannot solve.
By Sean K.S. Shin
This blog summarizes research trends based on published paper abstracts. Specific numbers or findings may contain inaccuracies. For scholarly rigor, always consult the original papers cited in each post.
The digital divide was first identified in the 1990s as a gap between those who had access to computers and the internet and those who did not. The assumption was that the divide would close as technology became cheaper, connectivity expanded, and digital literacy spread. Three decades later, the divide has not closed—it has transformed. The first-level divide (access) has narrowed in many regions but persists in rural areas, low-income communities, and the Global South. The second-level divide (skills and usage quality) has widened as the internet's role in employment, education, and civic life has expanded. And a third-level divide (AI readiness) is emerging as artificial intelligence reshapes which digital skills generate economic value.
The sociological significance of the digital divide lies not in the technology but in what it mediates. When employment applications move online, the digital divide becomes an employment divide. When education moves to platforms, the digital divide becomes an education divide. When government services go digital, the digital divide becomes a citizenship divide. The technology is the medium; the inequality is the message.
The Human Capital Connection
Liao, Kumar, and Furuoka (2025) investigate the relationship between the digital gap and the development of human capital concerning economic inequality. Despite the swift progression of the digital economy, which presents novel chances for knowledge acquisition and skill development, unequal access to the internet creates differential human capital accumulation that compounds over time.
The mechanism is cumulative: individuals with internet access develop digital skills, use those skills to access education and employment opportunities, and accumulate human capital that increases their economic productivity. Individuals without access fall further behind with each year, as the economy increasingly rewards digital competence. The digital divide thus functions as a multiplier of pre-existing inequality—converting a technological access gap into an accelerating human capital gap.
The Three-Level Divide
Baraka (2024) provides a comprehensive examination of digital divide and social inequality through a desk research methodology. The study synthesizes existing evidence on the multiple dimensions along which digital exclusion operates.
The analysis distinguishes three levels:
First-level divide (access): Physical access to devices and connectivity. This divide has narrowed globally—smartphone penetration exceeds 80% in high-income and many upper-middle-income countries. But access quality varies enormously: a student accessing a MOOC on a shared smartphone over an intermittent 2G connection has a fundamentally different experience than a student on a personal laptop with broadband.
Second-level divide (usage): The quality and diversity of internet use. Even among connected populations, usage patterns vary by education, income, and social capital. Higher-SES individuals are more likely to use the internet for education, career development, civic engagement, and financial management. Lower-SES individuals are more likely to use it primarily for entertainment and social communication. The divide is not just who is online but what they do there.
Third-level divide (outcomes): The economic and social returns to internet use. Even identical usage patterns produce different outcomes depending on the user's social position. A LinkedIn profile yields more employment leads for a user with an elite university credential than for a user with a community college background. Digital tools amplify existing advantage as readily as they create new opportunity.
AI and Automation: The Next Divide
Lahiri (2024) examines the sociological implications of the digital divide specifically in the context of AI and automation. With the rise of AI, digital technologies have become increasingly central to various aspects of life, creating new forms of exclusion for those who lack access.
The AI divide adds a dimension that previous digital divide analyses did not anticipate. As AI tools become integrated into professional work (AI-assisted coding, AI-augmented research, AI-driven decision-making), the ability to use AI effectively becomes a new axis of digital competence. Workers who can leverage AI to increase their productivity will capture disproportionate economic returns. Workers who cannot—whether because they lack access, training, or the foundational digital literacy that AI tools require—will face economic displacement.
The AI divide is particularly consequential because it may be self-reinforcing in ways that previous digital divides were not. AI tools improve with use; users who interact with AI systems regularly develop skills and workflows that make them more productive, generating more value, justifying more AI investment. Those excluded from this cycle fall further behind not linearly but exponentially.
The Socioeconomic Mechanisms
Kem (2024) examines the socioeconomic, regional, and demographic factors that drive digital exclusion. The digital gap is caused by several socioeconomic, regional, and demographic reasons, and it exacerbates existing inequalities across multiple dimensions.
The analysis identifies several reinforcing mechanisms:
- Income: Internet access and quality devices require recurring expenditure that low-income households cannot prioritize over food, housing, and healthcare.
- Geography: Rural and remote areas have lower connectivity infrastructure investment because per-capita returns to infrastructure providers are lower.
- Education: Digital literacy correlates with formal education, creating a chicken-and-egg problem: education builds digital skills, but increasingly, digital skills are needed to access education.
- Age: Older adults, who were not socialized with digital technologies, face higher barriers to adoption—relevant as populations age and digital government services replace in-person alternatives.
- Gender: In many Global South contexts, women have lower internet access than men, due to cultural norms, income disparities, and gender-differentiated access to devices.
Internet Access as a Human Right
Petkovic (2025) argues for recognizing internet access as a universal human right. In the era of digital transformation, internet access is no longer merely a technical issue but a question of human rights, equality, and social inclusion. The Internet today forms the foundation for access to education, healthcare, employment, political participation, and information.
The human rights framing shifts the debate from market provision (internet access as a commodity that consumers purchase) to public obligation (internet access as a right that governments must ensure). If internet access is a human right, then digital exclusion is a rights violation—with corresponding obligations on states to provide universal, affordable, quality connectivity.
Several international bodies have moved in this direction: the UN Human Rights Council resolution on internet access (2016) the ITU's Connect 2030 agenda, and the EU's Universal Service Directive. But recognition as a right has not translated into universal access, and the gap between rights discourse and infrastructure reality remains wide.
Claims and Evidence
<
| Claim | Evidence | Verdict |
|---|
| The digital divide is narrowing | Baraka (2024): first-level (access) narrowing; second-level (usage) and third-level (outcomes) persisting or widening | ⚠️ Uncertain (depends on which level) |
| Internet access drives human capital development | Liao et al. (2025): positive relationship between access and human capital accumulation | ✅ Supported |
| AI creates a new dimension of digital exclusion | Lahiri (2024): AI readiness as an emerging axis of digital inequality | ✅ Supported |
| Market mechanisms will close the digital divide | Kem (2024): socioeconomic, geographic, and demographic barriers persist despite falling technology costs | ❌ Refuted |
| Internet access should be recognized as a human right | Petkovic (2025): normative argument supported by international declarations; implementation gaps remain | ✅ Supported (normative) |
Open Questions
Can AI paradoxically help close the digital divide it threatens to widen? AI-powered low-bandwidth applications, voice interfaces for non-literate users, and automated translation for non-English speakers could make digital services more accessible—if designed with inclusion as a priority.What infrastructure model can deliver universal connectivity sustainably? Public investment, public-private partnerships, community networks, and satellite internet (Starlink, OneWeb) offer different models. Which produces the most equitable outcomes?How should digital skills education adapt to AI? If AI automates many current digital skills (data entry, basic coding, document formatting), what digital competencies will remain valuable, and how should education systems teach them?Does closing the digital divide require closing other divides first? If digital exclusion is a symptom of income, education, and geographic inequality, can it be addressed without addressing these root causes?Implications
The digital divide is a social inequality problem expressed through technological exclusion. Technology policy alone—expanding broadband, distributing devices, teaching digital skills—addresses the symptoms without addressing the causes. A comprehensive response requires coordinating digital policy with education policy, economic policy, and social protection policy to address the multiple reinforcing mechanisms that produce and sustain digital exclusion.
The digital divide was first identified in the 1990s as a gap between those who had access to computers and the internet and those who did not. The assumption was that the divide would close as technology became cheaper, connectivity expanded, and digital literacy spread. Three decades later, the divide has not closed—it has transformed. The first-level divide (access) has narrowed in many regions but persists in rural areas, low-income communities, and the Global South. The second-level divide (skills and usage quality) has widened as the internet's role in employment, education, and civic life has expanded. And a third-level divide (AI readiness) is emerging as artificial intelligence reshapes which digital skills generate economic value.
The sociological significance of the digital divide lies not in the technology but in what it mediates. When employment applications move online, the digital divide becomes an employment divide. When education moves to platforms, the digital divide becomes an education divide. When government services go digital, the digital divide becomes a citizenship divide. The technology is the medium; the inequality is the message.
The Human Capital Connection
Liao, Kumar, and Furuoka (2025) investigate the relationship between the digital gap and the development of human capital concerning economic inequality. Despite the swift progression of the digital economy, which presents novel chances for knowledge acquisition and skill development, unequal access to the internet creates differential human capital accumulation that compounds over time.
The mechanism is cumulative: individuals with internet access develop digital skills, use those skills to access education and employment opportunities, and accumulate human capital that increases their economic productivity. Individuals without access fall further behind with each year, as the economy increasingly rewards digital competence. The digital divide thus functions as a multiplier of pre-existing inequality—converting a technological access gap into an accelerating human capital gap.
The Three-Level Divide
Baraka (2024) provides a comprehensive examination of digital divide and social inequality through a desk research methodology. The study synthesizes existing evidence on the multiple dimensions along which digital exclusion operates.
The analysis distinguishes three levels:
First-level divide (access): Physical access to devices and connectivity. This divide has narrowed globally—smartphone penetration exceeds 80% in high-income and many upper-middle-income countries. But access quality varies enormously: a student accessing a MOOC on a shared smartphone over an intermittent 2G connection has a fundamentally different experience than a student on a personal laptop with broadband.
Second-level divide (usage): The quality and diversity of internet use. Even among connected populations, usage patterns vary by education, income, and social capital. Higher-SES individuals are more likely to use the internet for education, career development, civic engagement, and financial management. Lower-SES individuals are more likely to use it primarily for entertainment and social communication. The divide is not just who is online but what they do there.
Third-level divide (outcomes): The economic and social returns to internet use. Even identical usage patterns produce different outcomes depending on the user's social position. A LinkedIn profile yields more employment leads for a user with an elite university credential than for a user with a community college background. Digital tools amplify existing advantage as readily as they create new opportunity.
AI and Automation: The Next Divide
Lahiri (2024) examines the sociological implications of the digital divide specifically in the context of AI and automation. With the rise of AI, digital technologies have become increasingly central to various aspects of life, creating new forms of exclusion for those who lack access.
The AI divide adds a dimension that previous digital divide analyses did not anticipate. As AI tools become integrated into professional work (AI-assisted coding, AI-augmented research, AI-driven decision-making), the ability to use AI effectively becomes a new axis of digital competence. Workers who can leverage AI to increase their productivity will capture disproportionate economic returns. Workers who cannot—whether because they lack access, training, or the foundational digital literacy that AI tools require—will face economic displacement.
The AI divide is particularly consequential because it may be self-reinforcing in ways that previous digital divides were not. AI tools improve with use; users who interact with AI systems regularly develop skills and workflows that make them more productive, generating more value, justifying more AI investment. Those excluded from this cycle fall further behind not linearly but exponentially.
The Socioeconomic Mechanisms
Kem (2024) examines the socioeconomic, regional, and demographic factors that drive digital exclusion. The digital gap is caused by several socioeconomic, regional, and demographic reasons, and it exacerbates existing inequalities across multiple dimensions.
The analysis identifies several reinforcing mechanisms:
- Income: Internet access and quality devices require recurring expenditure that low-income households cannot prioritize over food, housing, and healthcare.
- Geography: Rural and remote areas have lower connectivity infrastructure investment because per-capita returns to infrastructure providers are lower.
- Education: Digital literacy correlates with formal education, creating a chicken-and-egg problem: education builds digital skills, but increasingly, digital skills are needed to access education.
- Age: Older adults, who were not socialized with digital technologies, face higher barriers to adoption—relevant as populations age and digital government services replace in-person alternatives.
- Gender: In many Global South contexts, women have lower internet access than men, due to cultural norms, income disparities, and gender-differentiated access to devices.
Internet Access as a Human Right
Petkovic (2025) argues for recognizing internet access as a universal human right. In the era of digital transformation, internet access is no longer merely a technical issue but a question of human rights, equality, and social inclusion. The Internet today forms the foundation for access to education, healthcare, employment, political participation, and information.
The human rights framing shifts the debate from market provision (internet access as a commodity that consumers purchase) to public obligation (internet access as a right that governments must ensure). If internet access is a human right, then digital exclusion is a rights violation—with corresponding obligations on states to provide universal, affordable, quality connectivity.
Several international bodies have moved in this direction: the UN Human Rights Council resolution on internet access (2016) the ITU's Connect 2030 agenda, and the EU's Universal Service Directive. But recognition as a right has not translated into universal access, and the gap between rights discourse and infrastructure reality remains wide.
Claims and Evidence
<
| Claim | Evidence | Verdict |
|---|
| The digital divide is narrowing | Baraka (2024): first-level (access) narrowing; second-level (usage) and third-level (outcomes) persisting or widening | ⚠️ Uncertain (depends on which level) |
| Internet access drives human capital development | Liao et al. (2025): positive relationship between access and human capital accumulation | ✅ Supported |
| AI creates a new dimension of digital exclusion | Lahiri (2024): AI readiness as an emerging axis of digital inequality | ✅ Supported |
| Market mechanisms will close the digital divide | Kem (2024): socioeconomic, geographic, and demographic barriers persist despite falling technology costs | ❌ Refuted |
| Internet access should be recognized as a human right | Petkovic (2025): normative argument supported by international declarations; implementation gaps remain | ✅ Supported (normative) |
Open Questions
Can AI paradoxically help close the digital divide it threatens to widen? AI-powered low-bandwidth applications, voice interfaces for non-literate users, and automated translation for non-English speakers could make digital services more accessible—if designed with inclusion as a priority.What infrastructure model can deliver universal connectivity sustainably? Public investment, public-private partnerships, community networks, and satellite internet (Starlink, OneWeb) offer different models. Which produces the most equitable outcomes?How should digital skills education adapt to AI? If AI automates many current digital skills (data entry, basic coding, document formatting), what digital competencies will remain valuable, and how should education systems teach them?Does closing the digital divide require closing other divides first? If digital exclusion is a symptom of income, education, and geographic inequality, can it be addressed without addressing these root causes?Implications
The digital divide is a social inequality problem expressed through technological exclusion. Technology policy alone—expanding broadband, distributing devices, teaching digital skills—addresses the symptoms without addressing the causes. A comprehensive response requires coordinating digital policy with education policy, economic policy, and social protection policy to address the multiple reinforcing mechanisms that produce and sustain digital exclusion.
References (5)
[1] Liao, J., Kumar, S., & Furuoka, F. (2025). Bridging the Digital Divide: How Internet Access Shapes Human Capital Development and Economic Inequality. e-Bangi, 22(2), 03.
[2] Baraka, K. (2024). Digital Divide and Social Inequality. International Journal of Humanities and Social Science, 2083.
[3] Kem, D. (2024). Digital Divide: Examining Socioeconomic Inequalities in Internet Access and Usage. IJFMR, 6(6), 30386.
[4] Lahiri, A. (2024). Sociological Implications of the Digital Divide: Exploring Access to Information and Social Inequality in the Age of Artificial Intelligence and Automation. RRIJM, 9(1), 019.
[5] Petkovic, A.D. (2025). The Right to Internet Access as a Universal Human Right. Science Journal, 4(4), 119.