If you hang out on tech Twitter or skim headlines about ChatGPT, Claude, or Gemini, it is easy to believe we are all rushing into the same AI-powered future. Tools that write code, draft emails, summarize research, or generate images are just a browser tab away. In theory, anyway.

In practice, a new kind of inequality is quietly being built into the foundations of the AI era. It is less about who owns a laptop and more about who has fast enough internet, the right data, the right skills, and a say in how these systems are built and used. That is the Digital Divide 2.0 — and it is already reshaping who benefits from AI, and who gets automated, monitored, or simply ignored.

You do not have to be a policymaker to care. Whether you are an employee, a parent, a student, or a local business owner, this divide affects your career options, your kids’ education, and the kinds of services your community will get in the next decade.

From “who’s online” to “who gets AI superpowers”

The original digital divide conversation was about access to the internet and devices: who had a connection and who did not. That gap is still very real. The World Bank estimates that about 2.6 billion people — roughly one third of the world’s population — remain offline, and internet use ranges from over 90% in high-income countries to just 27% in low-income countries (World Bank digital development overview).

But the debate has evolved. Researchers now distinguish between:

  • First-level divide: physical access (devices, connectivity).
  • Second-level divide: skills and digital literacy.
  • Third-level divide: whether people can actually turn technology use into better jobs, education, income, or health outcomes (summary of digital divide research).

AI pours gasoline on all three levels:

  • You need reliable, fast connectivity and modern devices to use advanced AI tools at all.
  • You need new skills — prompt design, critical thinking, AI literacy — to get value from tools like ChatGPT, Claude, or Gemini.
  • And you need institutional backing (supportive employers, schools, governments) to turn that usage into better opportunities instead of being replaced or sidelined.

That is Digital Divide 2.0: a shift from “who can get online” to “who can turn AI into actual power and progress.”

The global AI infrastructure gap

Behind every slick chatbot is a brutal amount of compute, data centers, and cloud infrastructure. That hardware is not evenly spread around the globe — and that matters.

The World Bank’s 2025 “Digital Progress and Trends” work on AI foundations shows that high-income countries dominate AI innovation, compute capacity, and startup funding, while many low- and middle-income countries struggle with basic cloud and data infrastructure and with retaining AI talent (World Bank Digital Progress and Trends 2025). Another World Bank note on digital and AI confirms that gaps in internet speed, data traffic, and digital use are hampering digital gains, especially in low- and middle-income economies (World Bank press release on uneven digital landscape).

Translated:

  • Countries that already had robust digital infrastructure are best positioned to build and deploy powerful AI.
  • Regions lacking cloud data centers, reliable power, and high-speed networks will be consumers of AI products built elsewhere, not producers.
  • That also means local languages, cultural contexts, and problems may be underrepresented in training data — and in the tools themselves.

So even if you can open ChatGPT in a browser, your country’s ability to create its own AI systems, train local models, or set terms with big tech is limited by these infrastructure gaps.

The AI skills divide at work

It is not just countries. Within workplaces, a new divide is emerging between people who know how to use generative AI well and those who do not.

McKinsey’s research on the state of AI in early 2024 found that organizations seeing the largest performance benefits from generative AI are concentrated in certain sectors (like tech, finance, and professional services) and tend to have more advanced digital capabilities, stronger data foundations, and dedicated AI training programs (McKinsey ‘State of AI in early 2024’ report). Another McKinsey analysis estimated that generative AI could add $2.6–$4.4 trillion in economic value annually across dozens of use cases, with the largest impact on knowledge work tasks like programming, marketing, and customer operations (McKinsey on the economic potential of generative AI).

That is great if:

  • You work in a company investing in AI tools and training.
  • Your job involves tasks AI is good at augmenting (writing, coding, analysis).
  • You have time and support to experiment with tools like ChatGPT or GitHub Copilot.

It is less great if:

  • You are in a smaller business or public sector organization without AI budgets.
  • Your role is routine, lower-paid, or not desk-based (many service and frontline jobs).
  • Your manager sees AI primarily as a way to cut headcount, not grow people.

An emerging pattern is that higher-wage, already well-educated workers are often the first to get AI “superpowers” — copilots for coding, communication, and research — while lower-wage workers are more likely to face intensified monitoring, automated scheduling, or replacement.

The risk: AI becomes a productivity booster for people who are already doing relatively well, while everyone else faces a mix of disruption and surveillance.

Education: AI as accelerator or amplifier of inequality

Education is another fault line. Generative AI could, in theory, help close learning gaps with personalized tutoring, translation, and support for students with disabilities. UNESCO’s Global Education Monitoring Report has highlighted how AI tools can provide tailored help for learners in remote areas or with special needs (UNESCO Courier on education in the age of AI).

But there is a catch: those benefits assume students and teachers have:

  • Reliable internet and suitable devices.
  • Safe, language-appropriate AI tools.
  • Training on how to use them ethically and effectively.

UNESCO’s 2023 guidance on generative AI in education warns that without strong policies on teacher training, data protection, and equity, AI in schools could deepen existing divides — benefiting affluent, well-resourced schools while under-resourced systems fall further behind (UNESCO guidance on generative AI in education and research).

So we get two very different classroom realities:

  • In well-funded schools: students using ChatGPT-style tools to brainstorm, get feedback, and explore ideas with guidance from trained teachers.
  • In under-resourced schools: spotty connectivity, outdated devices, and bans on AI tools because there is no capacity to integrate them responsibly.

That is the Digital Divide 2.0 in action — not just who has a laptop, but who learns to treat AI as a partner instead of a black box or a cheating tool.

Inside countries: race, class, disability, and rural gaps

Even in wealthy countries, AI access does not land evenly.

In the United States, for example, the older digital divide has not disappeared. Pew Research data cited in overviews of the U.S. digital divide notes that millions of Americans still lack home broadband and that access correlates strongly with income, race, age, and geography (summary of the U.S. digital divide). Rural communities frequently face slower speeds and fewer choices. Low-income households are more likely to rely on a single smartphone with limited data plans.

Layer AI on top of that and you get:

  • Students who cannot reliably access web-based AI tools to assist with homework.
  • Job seekers who cannot use AI to improve resumes, cover letters, or interview prep.
  • Small rural businesses that cannot afford or technically manage AI marketing, analytics, or automation tools.

There are also accessibility issues. Many AI tools are not yet optimized for people with disabilities, low digital literacy, or limited English proficiency. If user interfaces assume a certain reading level, cultural context, or jargon, some groups will find AI confusing or even hostile.

The result is that within a single city, some neighborhoods are effectively plugged into the AI economy, while others are left working with yesterday’s tools.

Who gets to shape AI — and whose data trains it?

One final, often overlooked layer of Digital Divide 2.0 is about voice and control.

AI models are trained on massive datasets scraped from the internet and other sources. If your language, culture, or community is underrepresented online, it is likely underrepresented in these datasets too. That can lead to:

  • Poor performance for certain languages or dialects.
  • Biased outputs that stereotype or misrepresent marginalized communities.
  • Tools that simply do not “see” problems specific to your context.

Meanwhile, the organizations setting AI norms — tech giants, major universities, and well-funded startups — are heavily concentrated in a handful of countries and cities. If your government, civil society, or local researchers are not at those tables, your interests may show up late or not at all.

So Digital Divide 2.0 is not just about users; it is about who gets to define what safe, useful, “responsible” AI looks like — and whose values and priorities are baked into the systems everyone else ends up using.

So what can you actually do?

This can sound abstract and intimidating, but there are concrete steps you can take to avoid being left behind — and to help close the gap for others.

  1. Invest in your own AI literacy, even in small bites.
    You do not need to become a machine learning engineer. Start with:

    • Trying tools like ChatGPT, Claude, or Gemini for everyday tasks (summarizing, drafting, brainstorming).
    • Practicing “prompting” — asking clear questions, giving examples, and checking answers critically.
    • Following a couple of reputable newsletters or courses focused on practical AI skills, not hype.

  2. Push your workplace or school toward inclusive AI use.
    Ask questions like:

    • “What training will we get on using AI tools safely and effectively?”
    • “How will we make sure these tools do not just help managers, but also frontline staff or students with fewer resources?”
    • “How are we checking for bias and errors in AI outputs before they affect people?”

  3. Support policies and projects that close infrastructure and skills gaps.
    That can look like:

    • Backing local efforts for better broadband and public access points (libraries, community centers).
    • Supporting programs that teach AI basics to underrepresented groups — not just kids in elite schools, but adults in career transitions, rural communities, and people with disabilities.
    • Paying attention to how your city, school district, or country is implementing AI in public services and education, and voicing concerns when plans ignore equity.

The AI revolution is not a single wave that either lifts everyone or crashes on everyone at once. It is more like a rising tide that reaches some shores faster, with stronger currents and better boats. Digital Divide 2.0 is about who gets those boats — and whether we are willing to build more of them, on purpose, before the gap becomes permanent.