✨ What we know about AI and the economy ... so far, at least

More investment and more business adoption ... with bigger impacts looming

Every quarter, Goldman Sachs gives a handy update on the (let’s hope) emerging Age of AI, tracking three key categories: AI investment, business adoption, and job market impact. The bank’s verdict through the third quarter of this year: “AI investment remains strong as adoption inches up.”

A few details:

✨ AI investment. Looks “strong,” especially for chip companies, with stock markets anticipating sustained growth through 2025. “While AI-related investment is not yet visible in national accounts data, manufacturers’ shipments for AI-related components remain elevated in the US and Japan.”

✨ Business adoption. Not many companies are actually using AI yet. Only a “modest” 6 percent of US companies use AI to make their products or provide services, up slightly from 4.6 percent earlier this year. But, but, but … “industry surveys released over the past quarter suggest that a large share of businesses are planning to increase investment in AI and related infrastructure, but many still have concerns over technological infrastructure and ethics and governance.”

✨ AI and jobs. AI is starting to affect jobs a little bit. There are more job openings for AI-related work compared to other jobs. The unemployment rate for jobs that could be affected by AI is slightly higher than for other jobs. In August, more companies mentioned AI when they announced layoffs, but it's still not a common reason for job cuts.

That final GS chart on labor productivity gives a feel for why AI enthusiasts are so, well, enthusiastic about how AI could transform the entire economy as the technology diffuses wide and deeper — and likely continues to improve. 

Which brings us to the General Purpose Technology Paradox, a counterintuitive historical phenomenon that challenges AI skeptics. Goes like this: The more important a new technology — steam power, electricity — turned out to be, the longer it took to realize its full potential. Big changes to industrial production played out over decades, not years.

The GPT Paradox deeply informs the new paper “Technological Disruption in the US Labor Market” by David Deming, Christopher Ong, and Lawrence Summers.