If you believe the headlines, artificial intelligence is on the verge of either gifting us utopia or unleashing our doom—and soon. The most optimistic voices in Silicon Valley speak of an imminent “economic explosion,” a singularity where super-intelligent machines automate everything and growth goes vertical. But is that the most likely outcome? According to a sobering new analysis from one of Stanford’s top economists, the most probable path for AI is far more gradual—a slow transformation that could take decades to fully materialize.
The “Explosion” vs. “Business as Usual” Scenarios
To understand the debate, Stanford Professor Charles “Chad” Jones presents two opposing futures. The “Explosion” view—championed by AI leaders like Sam Altman and Dario Amodei—envisions a world where AI masters all cognitive tasks and, when paired with advanced robotics, automates the physical world too. The result would be billions of super-intelligent virtual assistants, leading to unimaginable productivity and explosive economic growth.
The “Business as Usual” scenario, however, looks to the past. Over the last 150 years, revolutionary technologies like electricity, the automobile, the internet, and the semiconductor have each transformed society. And yet, through it all, U.S. living standards have grown at a remarkably steady 2 percent per year. This suggests that even a technology as profound as AI might simply extend our current growth trajectory, rather than radically upend it.
How “Weak Links” Bottleneck the Boom
So, which is it? Jones finds the answer in a simple principle: a chain is only as strong as its weakest link. In an economy built on countless interconnected tasks, a single bottleneck—one critical link in the chain—can throttle the entire system’s productivity.
What is that critical bottleneck? It’s us. As Jones astutely notes, “My computer can invert matrices like nobody’s business, but I have to figure out what data to put in those matrices, what questions to ask.” While our machines get exponentially more powerful, humans—the gatekeepers of creativity, problem-framing, and final judgment—remain stubbornly, biologically slow.
The economic data backs this up. The amount of computing power has exploded exponentially. Yet, the share of U.S. GDP paid as a return to computers is lower today than it was during the dot-com era of the late 1990s. In other words, computers have become super-abundant, and what is abundant commands little value. What remains scarce—human insight and the non-automated “weak links”—is what will continue to drive economic value.
The “Slow Singularity”: A 30-Year Horizon
This weak-links framework leads Jones to a provocative conclusion: even if AI becomes a profoundly transformative technology, its economic impact will be a slow burn. In his most aggressive calculations, it would take at least 30 years for AI to cause a genuine economic explosion. This is not because the technology is unimpressive, but because the process of diffusing it through the entire economy and, most importantly, automating the thousands of non-AI bottlenecks will be a long and arduous process.
This “slow singularity” concept re-frames the entire AI discussion. The real story isn’t about a sudden, apocalyptic shift, but about a decades-long, continuous effort to widen the circle of automated tasks. Growth, in this view, is not a one-time event but a constant struggle against new bottlenecks that will emerge even as old ones are eliminated.
In this gradual transformation, where innovation happens one task at a time, the true value won’t just be in owning the most advanced AI, but in strategically identifying and automating those persistent “weak links” that keep the rest of the chain from moving forward.
Source
| “A.I. and Our Economic Future,” Professor Chad Jones | “A.I. and Our Economic Future,” Professor Chad Jones – YouTube |
