The Growth Machine: AGI, Historical Hype, and Human Hurdles
Until about 1700, if you’d asked anyone about economic growth, you'd likely have received little more than a blank stare. For nearly two millennia, economies didn't so much grow as stumble in place—barely expanding at a snail's 0.1% per year. A Roman builder from 200 AD would have felt right at home working in 16th-century England—same techniques, same wages, same grimy hand tools.
Then something odd happened. Around the time Johann Sebastian Bach was scribbling fugues and Isaac Newton was explaining gravity, the spinning jenny appeared, steam engines chugged to life, and the economic growth engine began its slow, steady roar. Between 1700 and 1820, growth quintupled to a previously dizzying 0.5% per year. By the 19th century’s end—think Sherlock Holmes and telegraphs—it was up to 1.9%. During the 20th century’s rollercoaster ride, it accelerated again to 2.8%, meaning that every 25 years we produced twice as much stuff. Progress was no longer occasional, but relentless.
Now along come the digital-age alchemists of Silicon Valley with an even bolder claim: artificial general intelligence (AGI) will send growth soaring skyward again, this time to a mind-bending 20-30% annually. Sounds ludicrous, doesn't it? Well, remember, steam engines and telegraphs would have seemed equally preposterous to medieval peasants.
The logic here is beguilingly simple: historically, more people meant more ideas. More ideas meant more invention, productivity, and ultimately, prosperity. That's why Thomas Malthus, who worried that population growth would outpace food production, got things exactly backward. Instead of mass starvation, we got mass innovation. The bet with AGI is that the same kind of exponential increase can occur—this time without the messy business of having more mouths to feed.
But let's put on our skeptical hats for a moment. For starters, the most optimistic projections are built on models that resemble frictionless perpetual-motion machines more than real-world economies. As MIT's Daron Acemoglu points out, current evidence suggests only a small slice of tasks can be cheaply automated. But that assumption might soon be as obsolete as the spinning jenny itself—AI capabilities improve faster than early estimates tend to anticipate.
Yet here's the catch: historically, each revolutionary leap—from the spinning wheel to electricity—took longer to implement than anyone expected. Remember, Edison thought we'd all have electric homes within a decade. It took almost half a century. AI might automate entire workflows, sure, but the tricky bits—those stubborn human details, the political wrangling, the institutional inertia—these tend to slow even the slickest technologies to a crawl.
Moreover, for AI to truly turbocharge growth, it must tackle the toughest task imaginable: inventing better technology itself. If AGI can indeed automate research, achieving breakthroughs in energy, biotechnology, or even better AI, the feedback loop could accelerate until it approaches a singularity—a term economists borrowed from physics, meaning a point where growth becomes theoretically infinite. It's an idea so wild that even William Nordhaus, a Nobel laureate, admits it's probably impossible. Probably.
But even short of singularity fantasies, rapid automation might dramatically alter how society operates. History reminds us how tricky these transformations can be. Early industrialization was famously unkind to English workers—living standards stagnated or even declined between 1800 and 1850. Similarly, AI could depress wages, pushing earnings down until human labor becomes a quaint, inefficient curiosity.
Then again, as economist Tyler Cowen notes, bottlenecks always loom: human stupidity, resource constraints, political meddling, regulatory sluggishness. Even if AGI is astonishingly clever, it can't overcome energy shortages or rare-earth metal embargoes overnight. After all, your shiny AI-powered economy still needs electricity, chips, and stable politics—all notoriously temperamental ingredients.
What might emerge is a strangely uneven economy. Superstar workers who can harness AI might see explosive wage growth—think Renaissance master artisans in an age of guilds and guildmasters—but mundane jobs could vanish into a mist of automation. The result? An economic landscape resembling 18th-century aristocrats who paid extravagantly for services that couldn't yet be mechanized: childcare, fine dining, personal coaching.
Yet here we hit another historical snag—the so-called Baumol’s cost disease, named after economist William Baumol. Simply put, as some sectors automate rapidly, those that don't become prohibitively expensive. In the 20th century, food went from 43% of American incomes in 1909 to just 11% today, not because we eat dramatically less (we actually eat more), but because productivity exploded in agriculture. Meanwhile, prices ballooned in education, healthcare, and childcare—sectors stubbornly resistant to automation.
Finally, consider finance. High growth should mean high returns—but historically, explosive growth also brings volatile reactions: price controls, taxes, political instability. When inequality spikes, social unrest often follows, as history from Victorian factories to the streets of 1968 Paris clearly demonstrates. Betting on stocks or land, as AI evangelists suggest, might be riskier than it seems, especially if governments react with taxation or even nationalization.
So where does this leave us? AGI undoubtedly promises transformation akin to industrial revolutions of centuries past. But let's temper Silicon Valley's boundless optimism with historical caution. If we've learned anything, it's that technology alone isn't destiny. The tricky bit—the part history relentlessly reminds us—is how humans and their institutions choose to respond.
In other words, if this brave new AGI-powered world is coming, it won't be because a machine flipped a switch—it'll be because society decided, carefully and messily, to let it happen.
Info taken from this great article:
https://www.economist.com/briefing/2025/07/24/what-if-ai-made-the-worlds-economic-growth-explode?taid=6882724e0cf97000012d3ec9&utm_campaign=trueanthem&utm_medium=social&utm_source=twitter
Comments
Post a Comment