The Long-Term Effects of Generative AI on Productivity and Economic Growth
Generative artificial intelligence (AI) technologies, such as large language models, are increasingly transforming the landscape of work and productivity. These tools hold the potential to significantly impact economic output, with estimates suggesting a 15 percent increase in productivity and GDP by 2035, rising to nearly 30 percent by 2055, and reaching 37 percent by 2075. The largest boost is expected in the early 2030s, driven by rapid AI adoption across sectors.
One of the key findings highlights that approximately 40 percent of current gross domestic product (GDP) could be substantially affected by generative AI. This influence is particularly notable in occupations around the 80th percentile of earnings, where about half of their work content could be automated. Conversely, the highest and lowest earning occupations remain less exposed, suggesting an uneven impact across sectors and income groups.
AI's impact on productivity growth is projected to be most intense in the early 2030s, with a peak contribution of 0.2 percentage points in 2032. After this peak, the adoption and automation rates stabilize, causing growth to revert to its trend. Sectoral shifts, where more AI-exposed sectors experience faster productivity gains, contribute a lingering effect, raising aggregate growth by a permanent 0.04 percentage points. Consequently, total factor productivity (TFP) and GDP levels are forecast to be 15 percent higher by 2035, nearly 30 percent by 2055, and 37 percent higher by 2075, embedding a lasting expansion in economic activity.
These projections, however, should be interpreted cautiously. They rely on limited early data about AI's effects, and future technological advancements or unexpected developments could alter these estimates substantially. The real-world impact on productivity, employment, and public finances remains subject to ongoing research and wider adoption patterns.
The pace of AI diffusion is akin to historic technological shifts like the adoption of personal computers and the internet. Surveys indicate that by late 2024, around a third of workers and adults are already using generative AI tools, with adoption rates closely following patterns observed during earlier technological revolutions. As adoption accelerates, employment in occupations with high AI automation potential has begun to stagnate or decrease, especially among the most exposed jobs.
Occupational analysis reveals that occupations earning around the 80th percentile are most exposed, with approximately half of their tasks susceptible to automation by AI. Lower and higher wage occupations show less exposure, highlighting a widening gap in automation potential across income groups. The estimated 40 percent of labor income exposed to AI automation indicates substantial shifts in the workforce and economic activity, with about 23 percent of exposed tasks expected to be automated in the coming decades.
Cost savings from AI adoption, based on recent studies, average around 25 percent in labor costs, with future gains expected to rise to approximately 40 percent. These efficiencies are drivers behind faster productivity improvements and sectoral growth, especially in more exposed industries like professional services and technology.
Overall, while AI promises to enhance productivity significantly, its full macroeconomic impact will depend on the speed of adoption, technological progress, and sectoral transitions. The long-term outlook paints a picture of a larger, more productive economy with persistent gains in productivity levels, albeit with cautions regarding unforeseen shifts and challenges.