Productivity and Growth

Why are people in wealthy countries today able to afford things that were luxuries a century ago — or things that simply did not exist? The answer, at the deepest level, is productivity. Productivity is the ratio of output to input: how much a worker, a firm, or an economy can produce from a given amount of time, capital, and materials. When productivity rises, living standards rise with it. And AI is, in theory, one of the most powerful productivity tools humans have ever built.

The Mechanics of Productivity

Economists measure productivity in several ways. Labor productivity is output per hour worked — the most intuitive measure. Total factor productivity (TFP) is broader: it measures output relative to all inputs, including capital and labor together. TFP growth is often called the 'residual' — the portion of economic growth that cannot be explained by simply adding more workers or more machines. It reflects genuine efficiency improvements: better processes, better technology, better organization. Historically, TFP growth has averaged around 1-2 percent per year in advanced economies. That sounds small, but compounded over decades it is transformative. At 2 percent annual TFP growth, an economy doubles its productive capacity every 35 years. Economists believe AI could push TFP growth significantly higher, at least temporarily — some estimates from organizations like Goldman Sachs and McKinsey project AI could add 0.5 to 1.5 percentage points per year to TFP growth in advanced economies once adoption matures. Why? Because AI directly attacks the bottleneck that limits TFP: cognitive work. Physical capital (machines) replaced muscles; AI replaces and augments minds. If AI can accelerate research, reduce decision-making errors, automate knowledge-work bottlenecks, and compress the time from idea to implementation, it can raise TFP across virtually every industry.

The productivity channel AI is most likely to affect first is what economists call worker augmentation — AI tools that make a given worker substantially more productive without eliminating that worker's role. Evidence is emerging. A 2023 randomized controlled trial by Erik Brynjolfsson and colleagues at Stanford found that customer-service workers who had access to a generative AI assistant resolved 14 percent more customer issues per hour, with the largest gains (34 percent) among the lowest-skilled workers. Another 2023 study of software developers (GitHub Copilot) found a 56 percent speed increase on coding tasks. These numbers are striking, but they measure narrow task performance in controlled settings — not economy-wide productivity. The macro effects depend on how quickly adoption spreads, how well firms reorganize around AI capabilities, and whether the productivity gains stay inside leading firms or propagate through competition.

AI and the Innovation Process Itself

Beyond task-level augmentation, there is a more profound channel: AI accelerating the rate of scientific and technological innovation itself. Innovation is the engine of long-run growth in the models developed by economists Paul Romer and Robert Lucas. If AI speeds up the discovery of new drugs, materials, algorithms, and processes, it does not just help existing industries run more efficiently — it creates entirely new products that expand the economy's production possibility frontier. There are already concrete examples. DeepMind's AlphaFold 2, released in 2021, predicted the 3D structure of essentially every known protein — a problem biologists had worked on for 50 years. It has already accelerated drug discovery research in ways that will take years to fully materialize in clinical treatments and economic statistics. AI-assisted chip design tools at companies like Google and Nvidia are compressing the design cycle for next-generation hardware, which in turn enables faster AI — a potential self-reinforcing loop. If AI systematically compresses the time from scientific question to actionable result, the rate of innovation could rise substantially. Some economists, including Daron Acemoglu at MIT and Erik Brynjolfsson at Stanford, have published significantly different estimates of this effect — ranging from modest (Acemoglu: maybe 0.5 percent cumulative GDP gain over a decade) to transformative (Brynjolfsson: potentially several percentage points of annual growth). The honest answer is that there is genuine uncertainty among experts.