Engineer checking a robot.
getty
Artificial intelligence is making some occupations much more productive right now—but only a few so far. The pace of future economic growth makes a huge difference to today’s business decisions. Should a company prepare for stronger demand for its products? Or will its products or services be made obsolete by AI? Although nobody can be certain, new economic research clarifies the issues and suggests that growth will be only a little faster for a while—and then accelerate sharply.
A simple view of AI and economic growth divides production into easy tasks and hard tasks. For this analysis, “easy” means amenable to help from AI. “Hard” tasks are those that AI cannot help with, at least for a long time. Think about an electrical contractor wiring a new building. AI is good right now at estimating costs and determining the most efficient path for the conduit. But AI is not currently running the wires and connecting them. Some tasks are AI-easy, others are AI-hard.
AI-Easy And AI-Hard
Stanford economist Charles Jones has described research based on this concept in a recent paper. My assessment of the easy-hard split right now is that “white-collar” work is easily produced with AI, while “blue-collar” work is hard. The market is rife with AI “harnesses” which apply a large language model (such as ChatGPT, Claude or Gemini) to a particular business process in a particular industry. A harness might address electrical contractor estimating for constructing warehouses, or insurance claims processing for water damage to single family residences. These are very narrow tools, optimized for specific processes.
But work that is today performed by hands has been slower to automate. Robots are improving rapidly but have a long way to go.
Jones presents a very simplistic mathematical model of the economy that generates interesting results. Total output depends on using AI-easy work and AI-hard work. There is a little ability to substitute one kind of work for another, but only to a small degree. He then poses the question, if AI-easy work becomes abundant, virtually free, by how much does total output in the economy increase? Using plausible estimates about the underlying economy, he expects a 19% increase in total production.
In layman’s terms, if white-collar work can all be done without cost by AI, then total output only increases by 19% because blue-collar work becomes a bottleneck.
But over time, the split between AI-easy and AI-hard changes, with much more of the work being AI-easy. After a while, growth just soars, exceeding our imaginations. In the meantime—which could last years—economic growth will be nicely higher but not huge.
Super Growth Requires Robots
Today most usage of AI is in the white-collar domain. That will boost the overall economy. But the rapid growth of total production will require robots for most of the blue-collar work performed now. That includes robots roofing houses, lifting packages and cleaning restrooms.
Lifting objects is part of great robotics. But there’s a long way between lifting a ten-pound box and lifting a ten-pound baby
The Near-Term Economic Forecast: Faster, Not Huge
My best guess at the economic forecast for the next decade is that overall economic growth will be faster. We’ve been growing by about two percent per worker, inflation adjusted. That may rise to three percent this decade. And over time, the difference will be huge. Ten years at two percent makes the economy 22% larger; at three percent annual growth it would be 34% larger. That’s nice but it’s not the huge game-changer that AI will eventually get us to.
Be the first to comment