If an obscure 1980s paradox is any guide, AI may be about to hit a huge tipping point

There’s an old joke among economists that goes like this: “You can see the computer age everywhere but in the productivity statistics.” I didn’t say it was a funny joke. But when labor economist Robert Solow originally wrote those words in 1987, they were certainly true. Personal computers, corporate mainframes, and the first vestiges of the modern internet were all anyone could talk about. Yet productivity wasn’t budging. These whizzy technologies, in short, weren’t earning anyone any money. The phenomenon became known as Solow’s Paradox. Of course, we all know how that story ended. By the mid-1990s, productivity was on a tear, and tech was making lots of people fabulously wealthy. And (despite a subsequent crash and recovery), tech is now the linchpin of the modern economy. Today, AI is following a similar path. And new data suggests that a similarly massive productivity–and wealth–tipping point may be just around the corner. Old paradoxes Since generative AI surged into mainstream usage with the launch of ChatGPT in 2022, it has largely followed the same path that computers did in their infancy. The world can’t stop talking about LLMs and AGI. Yet as late as last year, even the buzziest of AI companies earned shockingly little. OpenAI, for example, had annualized revenue of around $20 billion as of the end of last year. For comparison, the pest control industry is about the same size, and the pizza industry is about two times bigger. The chasm between excitement and actual economic impact shows up in bigger datasets, too. A massive study published in February asked 6,000 business leaders how AI was impacting their operations. The answer? Not at all. While 63% of business leaders say they’ve adopted AI, 90% found it had no impact on their firm’s employment or productivity. Official stats tell largely the same story. A study from the Federal Reserve Bank of Saint Louis found that generative AI led to a 5.4% improvement in worker productivity–hardly the massive, wor