Chad Jones, a Stanford professor who has spent over 15 years researching economic growth, recently gave a lecture on how artificial intelligence will reshape our future. He laid out two opposing scenarios and used simple examples to show where the truth likely lies.
Two opposite worlds
Two contrasting views. Neither will probably play out in full — but understanding the gap between them matters.
Explosion: AI founders in Silicon Valley — Dario Amodei, Sam Altman, Demis Hassabis — have been saying this for a decade. And the process has begun. In November 2025, Anthropic gave its model — Claude Opus 4.5 — the same two-hour take-home exam it uses when hiring software engineers. The result: it scored higher than any human in history. “That was already seven months ago, and the models have only gotten better — we’re up to Opus 4.7 now, two generations later.”
If AI masters all coding soon, we point it at building even stronger AI. The result: billions of “virtual research assistants” — each running 100 times faster than us — designing new chips, discovering drugs, engineering robots. Dario Amodei called it “a country of geniuses in a data centre.” If all cognitive and physical tasks shift to machines, the economy explodes.
Business as usual: over the past 150 years, living standards in the US have grown at roughly the same rate — about 2 percent per year. During those 150 years, electricity, automobiles, airplanes, antibiotics, transistors, the internet and computers all entered our lives. Each one transformed the world — yet growth stayed at 2 percent.
Why? Every technology eventually runs out of steam. If we’d only had the steam engine and never discovered electricity, growth would have slowed. Each new technology doesn’t cause an explosion — it allows 2 percent growth to continue for another 50 years. And it takes decades for these technologies to fully diffuse through the economy.
Weak links — the key to everything
Jones used a simple principle to explain where the truth lies between these two scenarios: a chain is only as strong as its weakest link.
The same applies in business. Releasing a new iPhone requires design, sourcing, precision manufacturing, delivery, advertising — if any single one fails, the value of everything else drops to zero. The Space Shuttle Challenger exploded because of a $25 rubber O-ring failure.
“In my pocket, I have a computer with 100 million times the transistors than the equivalent of me had in the 1970s. But I’m not 100 million times more productive at research. Why not? 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.” Humans are the weak link. No matter how powerful the computer gets, human speed stays the same.
What does this look like in real numbers? Computers are everywhere, but what share of GDP is paid as a return to computing power? During the dot-com boom in the 1990s, that share rose and peaked at 4.5 percent in 2000. Since then, it has fallen by a third — to 3 percent. Computers multiplied, but their price dropped, and they now receive a smaller share of GDP. The weak link model predicts exactly this: computers are abundant, humans are scarce — and scarcity is what commands value.
What the model shows
We’ve been automating the economy for 200 years — textile looms, tractors, railways, computers. If AI is simply a continuation of that chain, what happens? Start with a simple question. If we automated all software — created infinite amounts of it — how much richer would we be? The answer: just 2 percent richer. Why only 2 percent? Because software accounts for roughly 2 percent of GDP. Weak links constrain everything else. Making one thing infinitely better isn’t enough — you need to automate every weak link, one after another.
The model showed three paths:
Purple path — machines do everything, including generating new ideas without human involvement. Labour’s share of income falls to zero, capital’s share rises to 100 percent. Growth explodes.
Green path — nearly everything is automated, but 3 percent of tasks remain human. Magnus Carlsen plays chess, Messi plays football — some things only humans do. Result: even if 97 percent improves infinitely, the 3 percent of human tasks bottlenecks the entire system. Labour’s share rises to 100 percent — humans are scarce, machines are abundant.
Blue path — the middle ground. Capital and labour shares stay stable — one-third, two-thirds. Growth slowly accelerates — from 2 percent to 2.3, to 2.6, to 3. Eventually it reaches 50 percent per year — but that takes centuries. By 2050, we’d be just 4 percent richer than without acceleration. By 2075 — 15 percent.
The three paths diverge dramatically 200 years from now. But for the next 75 years, you can barely tell which one you’re on. What if AI starts improving across the entire economy at Moore’s Law speed — 10 percent per year — starting today? By 2030, we’d be 50 percent richer than the baseline. By 2050, growth exceeds 25 percent per year. That’s a real explosion.
But even in this most aggressive scenario, the explosion takes 30 years. Full acceleration doesn’t arrive until 2060. Why? Weak links again. AI can handle cognitive tasks quickly, but in the real world — building robot grippers, reorganising factories, changing laws — everything moves at its own pace.
AI will transform the economy — that much is clear. But not tomorrow. Computers became 100 million times more powerful, yet we didn’t become 100 million times richer — because no matter how fast the computer gets, someone still has to figure out what to ask it. AI hits the same wall: it writes code, analyses data, generates answers — but where factories need building, laws need changing, and robot hands need working, the real world moves at its own speed.
“They’re saying everything changes in 3 to 5 years. Even in my most aggressive calculations, it takes at least 30 years for AI to truly cause an economic explosion” — Chad Jones.
This article is based on “A.I. and Our Economic Future,” a lecture by Chad Jones, Professor of Economics at Stanford Graduate School of Business.
