News
Anthropic: AI Boom Could Push Knowledge Workers' Wages Below Zero
A new Anthropic economic report finds that in an extreme AI growth scenario through 2030, the US economy would expand by more than 30 percent, but knowledge workers' real wages would fall and unemployment would exceed the peaks of the 2008 crisis and the pandemic.
Contents
Anthropic has published a report modeling how the growth of artificial intelligence could reshape the United States economy through 2030, and its conclusions are far more unsettling than the industry's usual assurances of rising productivity for everyone. In the most aggressive scenario, the gains from AI flow almost entirely to owners of capital, while knowledge workers see their real wages decline even as the overall economy grows explosively.
Three Scenarios Through 2030
The report, produced under Anthropic's Economic Futures program by economists Anton Korinek of the University of Virginia, Charles I. Jones of Stanford, and Szymon Sacher, lays out three AI development paths that differ in how much task automation they involve. In the modest scenario, AI automates just 4 percent of tasks in the economy, GDP rises 1.6 percent above the baseline path, and unemployment ticks up slightly to 3.9 percent.
The scenario called 'substantial' assumes AI takes over 12 percent of tasks, mostly by replacing workers rather than augmenting them. GDP then rises 8.3 percent above baseline at an annual rate of 5.4 percent, employment in cognitive occupations falls by about 4 percent, and white-collar unemployment reaches 4.5 percent. This is the variant that most closely matches ordinary Americans' expectations.
The Extreme Scenario and the Split of Gains
The most alarming is the third, extreme scenario, in which self-improving AI performs nearly a third of all tasks in the economy. Annual GDP growth then reaches 15.4 percent, meaning the economy would double every 4.5 years, with GDP rising 32.4 percent above the baseline level. At the same time, overall unemployment stands at 11.9 percent, and 17.9 percent among white-collar workers, exceeding the levels recorded during the 2008 financial crisis and the coronavirus pandemic.
The key finding, however, concerns not the growth itself but how it is distributed. The labor share of GDP shrinks from 60 to 45.2 percent, while capital's share rises to 54.8 percent. This creates a paradox: even though the economic pie grows by a third, the total wage bill for workers ends up almost identical to a world without transformative AI. Real wages for knowledge workers fall 11.5 percent below the reference path.
The Bill for Offsetting the Losses
According to the report's authors, compensating knowledge workers for these losses would require massive social transfers on the order of 9 percent of GDP, a sum comparable to total US spending on Social Security and Medicare combined. That illustrates the scale of the fiscal challenge a scenario of rapid, self-improving automation would create if it actually materialized.
Anthropic compared its scenarios with a Morning Consult survey of nearly 11,000 Americans. The median of respondents' expectations roughly matches the 'substantial' scenario, with projected GDP growth of 8.6 to 10 percent and unemployment of around 4.6 to 5 percent by 2030. That suggests the public expects a moderate AI impact, far milder than the extreme variant described in the report.
Policy Proposals and Open Questions
Alongside the report, the company published a set of policy recommendations broken down by scenario. For all variants, it proposes worker training grants of around $10,000 a year, tax breaks that favor retraining over layoffs, and faster permitting for AI infrastructure. For the moderate scenarios, it suggests assistance programs modeled on Trade Adjustment Assistance along with taxes on computing power, and for the rapid-automation scenarios, sovereign wealth funds holding stakes in AI, a VAT, and taxes on corporate wealth.
We are uncertain, and so are outside experts - Anthropic, report on AI economic policy
The company itself acknowledges in the document that it cannot precisely determine the scale or pace of the economic transformation AI will bring, which is why it chose to present a range of scenarios rather than a single forecast. Anthropic also announced $10 million to expand the Economic Futures program, which will fund further empirical research into AI's impact on the economy.
What It Means for Poland
The report deals directly with the US economy, but its underlying mechanism, a shift of national income from labor to capital, is universal and applies to any economy built on knowledge work, including Poland's. Polish economists and institutions have warned for months that the country is among the most exposed in the European Union to the effects of office-job automation, and Anthropic's scenarios provide concrete figures that could be used to estimate similar effects locally.
For Polish businesses and policymakers, the key takeaway is the scale of social transfers that would be needed to cushion the effects of automation. If even the world's richest economy would have to spend the equivalent of its Social Security and Medicare budgets combined to compensate knowledge workers for their losses, the question of financing similar mechanisms in Poland becomes more urgent than the general debates over the AI Act and digitalization strategies have so far suggested.
