Anthropic models three AI futures: boom, disruption, or both

Anthropic has released economic modeling that maps three distinct trajectories for AI's impact through 2030, ranging from modest productivity gains to a transformation without historical parallel.

The research, developed by Anthropic's economics team, projects outcomes spanning from AI handling 4% of economy-wide tasks to nearly one-third of all work. The scenarios diverge sharply in their implications for growth, employment, and income distribution.

Why it matters

These projections frame a critical policy question: how societies will distribute gains from potentially unprecedented economic growth when that growth comes alongside significant worker displacement. With labor's share of national income already at historic lows, the research suggests AI could accelerate a shift of economic rewards away from workers and toward capital owners.

Three paths forward

In Anthropic's "modest" scenario, AI adoption remains limited. The technology handles just 4% of tasks by 2030, producing GDP growth 1.6% above baseline with minimal employment effects.

The "substantial" scenario envisions AI performing 12% of tasks, primarily replacing rather than augmenting workers. GDP expands 8.3% beyond the no-AI baseline, while unemployment climbs modestly to 4.6%.

The "extreme" scenario projects AI executing nearly 30% of economic tasks, mostly through worker replacement. GDP surges 32% higher by 2030, but unemployment approaches 12%.

White-collar workers face concentrated risk

Job losses concentrate heavily in cognitive work. Under the substantial scenario, white-collar employment contracts roughly 4%. In the extreme case, it falls more than 20%.

The income distribution effects prove equally stark. In the most aggressive adoption scenario, labor's share of national income drops from 60% to 45% by 2030. Despite the economy growing one-third larger, workers collectively earn no more than they would in a world without AI.

What determines the outcome

"These scenarios are not predetermined — it's not like an inexorable march," says Peter McCrory, an economist at Anthropic. The next one to two years should reveal which trajectory is materializing, according to Anton Korinek, another Anthropic economist, though other economic shocks could obscure early signals.

Key variables include the pace of capability improvements, how quickly AI diffuses across sectors, and the magnitude of productivity gains in affected industries.

Limitations and exclusions

The scenarios omit several important factors. They don't model political resistance such as opposition to data center construction or calls to pause AI development. Catastrophic risks from the technology itself fall outside the analysis, despite warnings from Anthropic researchers about existential threats.

The projections also exclude recessions, financial crises, and the possibility that AI-driven job losses could weaken consumer spending enough to trigger broader economic contraction. Advances in robotics that might automate physical labor aren't incorporated.

Anthropic provides an interactive tool allowing users to test custom assumptions across the modeling framework.

The research and projections were first reported by Axios.