🤖 If the internet was the ultimate tool for ideological polarization, AI threatens to cement a grimmer, more tangible polarization of social class. As we stand at the fresh beginnings of AI development, predictions point to a future of exacerbated inequality — a divide that may grow starker than ever.

Just as we’ve long been defined by income, education, and employment — and more recently our access to technology — our access to AI is poised to become a (if not the) social class marker. In an interview with the Financial Times, Nicolai Tangen, head of Norway’s USD 2 tn national fund — the world’s largest sovereign wealth fund — warned that AI has the potential to widen the gap between social classes across the globe.

“There is a potential for splitting societies, and there is a real potential for splitting the world into the counties which can afford it and the countries which cannot,” Tangen told FT. Over 85% of countries lack national AI strategies, a striking number that implies an already widening access gap, according to ITU Secretary-General Doreen Bogdan-Martin. She has also noted that with 2.6bn peoplestill offline, access remains a fundamental barrier: without addressing the digital divide, AI risks reinforcing global inequities.

The difference has already started manifesting in regulatory approaches. Tangen highlights the US’s rapid embrace of AI in contrast to the EU’s heavier regulation and slower adoption, arguing that the EU’s tendency to over-regulate may hold back economic growth. A study by the European Commission revealed that only 13.5% of European enterprises were using AI technologies as of 2024, compared to the US, where private AI investment reached USD 109 bn last year — nearly 12 times China’s, and 24 times the UK’s.

Multiple factors will determine a country’s place in the race. “You need prior education, you need electricity, you need digital infrastructure… There is a potential for this to amplify differences in the world,” Tangen said. Even with AI’s productivity gains — Tangen estimates as much as 20% in his own organization — policymakers risk falling behind the pace of technological change, though perhaps this will protect them once the AI bubble pops.

But the AI divide won’t just play out on the geopolitical stage — it will cut through workplaces, households, and individual careers. Access to advanced AI tools is stratified by cost — while basic chatbots are available for no charge, the most powerful AI capabilities are locked behind premium tiers. OpenAI’s ChatGPT Pro, for instance, costs USD 200 per month, offering access to advanced reasoning, extended research capabilities, and priority speed that non-paying users cannot utilize.

The implications for entry-level workers are particularly dire. Anthropic CEO Dario Amodei has issued one of the most direct warnings from inside the industry: AI could eliminate half of all entry-level white-collar jobs within five years. This threatens to disrupt the traditional career pipeline — if new workers cannot land entry-level roles, they won’t develop workplace skills — with severe consequences in the future, resulting in a generation locked out of the professional class before they even begin.

The productivity and income gap this creates may be profound — research from Brookings found that exposure to AI-driven productivity gains is concentrated at the higher end of the income distribution. Workers in lower-paid, in-person service and manual labor roles — the very people with the least disposable income to invest in AI tools — are the least likely to benefit from the technology’s effects. If AI significantly complements higher-income workers, it could lead to a disproportionate increase in their labor income, according to the IMF. “In most scenarios,” the IMF concluded, “AI will likely worsen overall inequality, a troubling trend that policymakers must proactively address to prevent the technology from further stoking social tensions.”

In a future dominated by AI automation, access to the technology will become the new wealth standard on both a national and individual level. Governments and large corporations will have to grapple with the consequences of uneven adoption: labor market disruption and, most significantly, questions of access and fairness. As access to advanced Ai models becomes increasingly expensive, capital will shift toward adoption — at the cost of humans, if history is any indicator. Those who can afford it will increasingly rely on AI computing instead of human labor — everything from software development to creative marketing to running factories will find more efficient automated alternatives.

A ‘permanent underclass.’ This prediction amplifies long-felt fears that AI threatens employability, leading to a “permanent underclass” — a concept explored by Kyle Chayka in the New Yorker. Chayka writes that the idea has been embraced online partly as a joke, and partly out of sincere fear about how AI automation will upend the labor market and create a new norm of inequality. As he writes, “fears of a permanent underclass reflect the fact that there is not yet a coherent vision for how a future AI-dominated society will be structured.”

That scenario isn’t as distant as it might seem. A speculative forecast of AI, developed by a team of researchers including former OpenAI employee Daniel Kokotajlo, predicts that artificial general intelligence (AGI) could arrive by 2027, with artificial superintelligence following months later. While predictions vary — some experts have pushed back on the timeline — the scenario envisions a point at which AI becomes self-reinforcing, automating its own development in an independent feedback loop. The CEOs of OpenAI, Google DeepMind, and Anthropic have all predicted that AGI will arrive within the next five years.

“We live in a time where it’s totally futile to try to predict anything,” Tangen acknowledges. “The focus now [is] on agility, culture, and preparing societies for what’s coming.” Whether or not such timelines prove accurate, they underscore an urgent question: as AI accelerates, who will be left behind?

Tags: