More than one in five white-collar workers could be unemployed within four years, the Anthropic Institute's research arm says.
In a report called Economic Scenarios for Transformative AI, researchers at the Anthropic Institute have mapped out how varying levels of artificial intelligence capability and adoption could affect the US economy, including GDP, wages and unemployment.
When it comes to unemployment among white-collar or “knowledge” workers, the report outlines three potential scenarios.
Under the modest scenario, AI continues to improve but substitutes for cognitive work only in limited ways. Disruption to the labour market remains relatively small, with unemployment reaching 2.9 per cent by 2030.
Under the major scenario, capabilities and adoption advance more sharply, allowing AI to perform far more cognitive work and displace a larger number of workers. Unemployment rises to 4.5 per cent.
But under the extreme scenario, capabilities advance exceptionally quickly and AI can substitute for a very large amount of cognitive work, substantially reshaping demand for human labour. Under these assumptions, unemployment among white-collar workers could reach 17.9 per cent by 2030.
That would mean almost one in five workers in management, professional, technical, sales and office-based occupations would be unemployed under the model’s most disruptive scenario.
AI could create a bigger economic pie, but who gets more of it?
AI could make the US economy much bigger over the next five years. But that does not necessarily mean workers will receive a bigger share of the wealth it creates.
There is intense debate about how much AI will reshape the global economy and labour market in the coming years. It is one of the biggest open questions in economics today. In an attempt to explore some of the possible scenarios, Anthropic has released a model mapping the best and worst-case scenarios for how AI could affect the US economy. The tool, released alongside a technical report titled Economic Scenarios for Transformative AI, models how AI could reshape jobs, economic growth, wages and unemployment through 2030.
One of the most striking findings is what could happen to the balance between workers and capital. Today, for every dollar the US economy produces, about 60 cents goes to workers and 40 cents to capital.
Anthropic looked at what could happen to that split as AI becomes more capable and businesses use it to perform more tasks. In its most modest scenario, not much changes. Workers still receive 59.4 per cent of economic output, compared with 40.6 per cent for capital. But as AI plays a bigger role in the economy, the gap starts to close.
In Anthropic's substantial scenario, labour's share falls to 56.1 per cent, while capital's rises to 43.9 per cent. And then, in the most extreme scenario, they switch places. The US economy grows to $44.4 trillion, but workers receive only 45.2 per cent of it. Capital takes the majority, at 54.8 per cent. That's a 14.8 percentage-point shift towards capital.
