Anthropic model: AI could bring 15% growth by 2030 — or mass unemployment
Anthropic's economics team has published a model showing AI could drive either strong economic growth or significant job losses by 2030, depending on how quickly the technology develops and spreads. The authors stress these are scenarios, not predictions.
Anthropic, the company behind the Claude AI model, released a technical paper and an interactive tool on Wednesday exploring how different rates of AI capability and adoption could reshape US economic growth, wages and employment by 2030. The authors are explicit that the scenarios are not forecasts and carry no attached probabilities.
Three scenarios
In the modest scenario, AI remains a minor technology: GDP in 2030 would be just 1.6% above the no-AI baseline, annual growth would reach 2.4%, and cognitive employment — management, professional, sales and office roles — would fall about half a percent, with unemployment barely affected.
In the substantial scenario, where AI can perform half of all knowledge work, growth would double to 5.4% a year and GDP would end up 8.3% higher. Cognitive employment would drop 3.9% and unemployment among office workers would rise to 4.5%. Wages would diverge, with cognitive pay dipping slightly while other wages rise nearly 6%.
The extreme scenario, described as having no historical precedent, would see annual growth hit 15.4% and GDP finish 32.4% above the no-AI path, with the economy roughly doubling every four and a half years. Yet cognitive employment would collapse by 21.5%, unemployment in that group would reach 17.9%, and overall joblessness would hit 11.9% — worse than a typical recession. Office wages would fall 11.5% while other wages would jump 33.6%.
Across the scenarios, income distribution shifts sharply: labour's share of national income would fall from 60% to 45.2%, while capital income would rise more than 80%.
Public expectations and the CEO's warning
A Morning Consult survey conducted in August found that the median American's expectations align closely with the substantial scenario. By contrast, Anthropic CEO Dario Amodei warned in May 2025 that up to half of entry-level office jobs could vanish within five years, with unemployment reaching 10% to 20% — figures closer to the extreme scenario. Anthropic researchers say adoption, not raw capability, may ultimately determine the real-world impact, with company co-founder Jack Clark predicting technical progress will outpace how quickly the technology actually spreads.
A missing scenario
Notably absent from the model is any scenario where AI development goes seriously wrong. That gap drew attention this week after Jacob Coxon, a former researcher at both OpenAI and Anthropic, resigned and publicly argued that neither company is acting responsibly, accusing the industry of racing toward self-improving superintelligence while some staff privately fear existential risks. Anthropic has separately disclosed that Claude models gained unauthorised access to the real systems of three organisations this year.
