Anthropic has set out three sharply different scenarios for the US economy by 2030, ranging from an internet-style productivity boost to a transformation in which artificial intelligence drives annual growth of 15 per cent but pushes unemployment to levels well beyond those seen in a typical recession.
The research, published alongside an interactive economic model, is intended to show how different assumptions about the capability, adoption and autonomy of AI could affect GDP, wages and employment. Anthropic stressed that the scenarios are not forecasts and that the future will depend heavily on how quickly businesses and workers adopt the technology.
In the most modest scenario, AI is used for about 4 per cent of economic tasks by 2030. Its effect is comparable to that of the internet: growth is stronger, but remains within the range associated with previous technological advances. US GDP would be 1.6 per cent higher than in an equivalent economy without AI, with little change in unemployment.
The substantial scenario assumes AI can perform half of all knowledge-work tasks by the end of the decade, with most of that work carried out autonomously. Adoption remains incomplete, however, and the technology is used for only about 12 per cent of tasks across the economy. GDP reaches $36.3 trillion, or 8.3 per cent above the no-AI baseline, while unemployment rises to about 4.6 per cent.
Knowledge workers would see little increase in pay in that scenario, while workers in less AI-exposed occupations could benefit from higher demand. Anthropic’s model suggests that coders and call-centre staff, for example, might increasingly move into roles such as nursing or electrical work, although switching occupations could leave many people unemployed for extended periods.
The most extreme scenario assumes AI becomes more productive than people at almost all knowledge-work tasks, performs nearly all of them autonomously and creates few new tasks for human workers. Annual GDP growth reaches 15 per cent, causing the economy to double in size roughly every four and a half years.
That surge would take US GDP to about $44.4 trillion by 2030, 32.4 per cent above the no-AI path. But overall unemployment would approach 12 per cent, while unemployment among knowledge workers would rise to historically high levels. Wages in those occupations would fall by more than 10 per cent, according to the model.
AI growth could widen the divide between workers and investors
Anthropic’s figures suggest that a larger economy would not necessarily mean a proportionately larger share for employees. In the extreme scenario, the share of national income going to labour falls from about 60 per cent to 45.2 per cent, while the share going to capital rises to 54.8 per cent.
The company said the result would be an economy that was substantially richer overall, but in which many knowledge workers faced lower wages or joblessness. The central policy challenge would therefore be ensuring that the gains were widely shared rather than concentrated among the owners of the systems, infrastructure and businesses benefiting from them.
An accompanying survey of 10,980 Americans found that the typical respondent’s expectations were broadly consistent with the substantial scenario. About one in ten respondents held views closer to the extreme outcome, while concerns about the effect on younger workers and entry-level jobs were particularly pronounced.
Anthropic’s model represents jobs as collections of tasks that can be left unchanged, assisted by AI, automated or replaced by new tasks. It uses occupational data from the US Department of Labour and allows users to alter assumptions about productivity, adoption, autonomy and the time workers take to find new employment.
The company cautioned that the tool is a simplified framework rather than a complete economic forecast. It does not model government policy responses, recessions, financial-market disruption, the effect of weaker consumer demand after job losses, or the possibility of AI systems causing catastrophic harm. It also excludes advances in robotics that could affect physical jobs.
Anthropic’s economists said the next year or two could provide early indications of which path is emerging, as researchers and businesses gauge how quickly AI capabilities improve and how widely they spread beyond the technology sector. The company’s own research acknowledges that increasingly capable systems could still have a limited economic effect if adoption remains slow.
