How could AI reshape the economy by 2030? Anthropic outlines 3 scenarios

Anthropic has sought to answer one of the most consequential economic questions of the AI era: how dramatically will artificial intelligence reshape the economy?

New research from the AI company outlines three potential paths through 2030, ranging from a relatively modest technological shift similar to the internet's early economic impact to an extreme scenario in which AI rapidly transforms productivity, employment and the distribution of income.

The company does not predict which scenario will ultimately materialize.

"The scenarios are not predictions, and we attach no probabilities to them; their purpose is to make the consequences of different assumptions comparable," Anton Korinek, an economist at Anthropic, said.

Instead, its economists argue that the next one or two years could provide some of the first meaningful clues.

"We will see in the next year or two how rapidly the capabilities increase, how rapidly the technology diffuses through the economy, and how much the productivity gains in all the sectors that it touches are," Korinek said.

Three possible paths for AI

Anthropic's scenarios are designed around different assumptions about how quickly AI capabilities improve, how widely the technology is adopted, and how much it boosts productivity on the tasks it affects.

The most cautious is the modest-change scenario.

Here, AI remains a relatively small technology, producing economic gains but largely keeping the economy on its existing trajectory.

By 2030, just 4% of economic tasks would have been affected by AI.

Productivity on those tasks would rise by around 35%, with half of affected instances automated and half augmented.

For every two tasks automated by AI, one new task would be created and performed by cognitive labor.

Under this scenario, GDP would be 1.6% above its no-AI path in 2030, equivalent to roughly 10 months of additional normal economic growth accumulated over four years.

Annual growth would reach 2.4%, compared with 2% without AI.

The labor-market effects would be limited.

Average wages would be 0.7% higher, while unemployment would stand at 3.9%, only slightly above the 3.8% rate assumed on the normal path.

The substantial-change scenario

The middle scenario envisages AI becoming a pivotal technology with an economic impact greater than that of the internet.

By 2030, AI would affect 12% of all economic tasks, equivalent to roughly 20% of the tasks performed by cognitive workers in 2025.

Productivity on affected tasks would increase by approximately 57%.

Three-quarters of those instances would be automated, while one-quarter would be augmented.

Unlike the modest scenario, where new tasks partially offset automation, the reallocation of labor becomes much more significant.

For every four tasks automated, one new task would be introduced and performed by cognitive workers.

The macroeconomic consequences would also be considerably larger. GDP would be 8.3% above its no-AI path by 2030, with annual growth accelerating to 5.4%.

That would be a striking increase by historical standards.

During the 1990s dot-com boom, the fastest annual US GDP growth rate was 4.7% in 1999.

The scenario also assumes significant changes in employment as workers move between occupations and industries.

The extreme scenario

At the other end of Anthropic's framework is an outcome in which AI fundamentally transforms the economy.

In the extreme scenario, AI would affect 30% of economic tasks by 2030, equivalent to roughly half of the tasks performed by cognitive workers in 2025.

AI would more than double productivity on the tasks it affects, while 90% of affected instances would be automated and only 10% augmented.

The scenario also removes the assumption that automation will generate enough new cognitive tasks to offset some of the jobs displaced.

The resulting economic acceleration would be unprecedented.

By 2030, GDP would be 32% above the no-AI path and 40% higher than in mid-2026. Annual GDP growth would reach 15.4%.

If such a growth rate were sustained, per-capita incomes would double roughly every five years.

In comparison, US per-capita incomes have historically doubled approximately every 35 years.

But the gains would come with substantial disruption to the labor market.

AI could reshape wages and income

The extreme scenario projects average wages 9.7% above the no-AI path.

However, the headline wage increase masks a major shift in how economic output is distributed.

Anthropic estimates that 15% of GDP could move from being paid to labor toward returns on investment.

That would mean the benefits of AI-driven productivity would not necessarily be distributed evenly between workers and owners of capital.

Unemployment would also rise sharply, reaching 11.9% across the economy.

That would be substantially higher than the peaks seen during recent US downturns.

Annual unemployment reached just below 10% after the global financial crisis and around 8% during the Covid-19 recession.

The scenario therefore illustrates the central tension surrounding AI: the technology could produce extraordinary increases in economic output while simultaneously creating severe disruption for workers whose tasks become automatable.

What Americans expect from AI

Anthropic also surveyed more than 10,000 Americans in August to gauge how people view current and future AI capabilities, adoption, and the ease of finding new employment if they are forced to change occupations.

The responses generally pointed toward an outcome between the modest and extreme scenarios.

"The typical respondent’s answers imply outcomes close to the “substantial change” scenario: GDP is 10% higher by 2030 than it would be without AI, and the overall unemployment rate has risen to around 5%. Around 10% of respondents have views in line with the extreme scenario," Anthropic said.

That suggests the public is already anticipating a meaningful economic transformation, even if relatively few respondents expect the most disruptive outcome.

Economists remain more cautious

Anthropic's researchers stress that the scenarios should not be interpreted as a forecast.

The uncertainty reflects the difficulty of determining how quickly AI systems will improve and how readily businesses will incorporate them into real-world workflows.

AI experts generally expect the technology to spread more rapidly, according to Anthropic, while economists tend to take a more cautious view.