Since the dawn of the industrial revolution, technology has played a trajectory-changing role in enabling human productivity, economic growth, and job creation. The inventions of technologies like the electric motor in 1890 and the personal computer in 1981 led to significant productivity booms in the early 1920s and early 2000s, respectively. These types of revolutionary developments catapult societies forward. A recent Goldman Sachs study found that 60% of workers today are employed in occupations that did not exist in 1940, implying that over 85% of job growth since the end of WW2 is technology-driven. When we leverage technology efficiently, we open ourselves up to a whole new productivity frontier.
Artificial Intelligence is the next electric motor, the next personal computer. And like the technologies that came before it, AI will be the driver of the next productivity boom. However, the fate of this boom will be different — it will reflect the novel societal and economic conditions we face over the next 20 years. Today is unlike the turns of the 20th and 21st centuries. Today, society faces accelerating challenges like population aging, low fertility rates, and productivity decline. Linear per capita productivity expansion will not be enough to sustain human economic production, we’ll need our productivity to increase exponentially. In this new world, it’s become clear that AI is not an accretive innovation; rather, it is a foundational innovation that humanity needs to embrace quickly if we are to overcome a potential economic collision course in the next 20 years.
Populations are growing older and less productive
Economic growth has always been driven by two key vectors: population increases (which increases the supply of labor) and increases in per capita productivity. During childhood and old age we are net consumers — we need more than we’re able to produce with our own labor, so we rely on others to take care of us. The difference between consumption and production in childhood and old age is made up partly by savings, but also largely by intergenerational transfers. Changes in fertility rates disrupt the delicate balance of these intergenerational economic dependencies. Falling fertility rates drive continued population aging via this economic feedback loop. For centuries, we’ve never had to think about this — the global population has been growing explosively, at times even too quickly. But now, we find ourselves confronted with a different predicament: what happens when our population starts to shrink? In 2012, a report by the United Nations found that 48% of the world lives in countries where the TFR (total fertility rate) is below replacement (~2.1 births per woman).
Changing population structures pose a direct threat not just to the social structures of our communities, but to the very fabric of our economic activity. Gross productivity is directly tied to the number of laborers in a given system — more workers equates to more output. As the global population ages, per-capita increases in productivity are necessitated in order to maintain (let alone expand) our current economic growth rate.
At the same time that populations are shrinking, issues with net output are exacerbated by falling rates of productivity per worker. According to NPR, at the end of 2022 US productivity was down 4.1% on an annualized basis, the most significant decline since the government started keeping track of the number back in 1948. With an aging population, a rapidly-declining ability to replace that population, and declining productivity, we’re not just overdue for a technological leap to fill the gap — we’re highly dependent on it. If we aspire to live in a future with positive economic growth, we will need to find a way to leverage AI to make everyone exponentially more productive — and fast.
The plausibility of exponential AI-driven increases in productivity
Robert Solow, a prominent American economist, proved the economic plausibility of using technology to improve quality of life back in the 1950s. In economic models that predate Solow (a famous example of this is Malthus and his dismal conclusions), economists believed that there was a finite level of output for society and that humanity was doomed to struggle to make do with this set output forever. We simply move along this one set output curve forever, doing a little bit better when the population shrinks and doing a little bit worse when the population grows. Solow discovered that in fact, we can see output as a function of technology. By developing new technologies, we can shift the whole output curve up. Best of all, according to Solow, innovations in technology increase the standard of living for everybody. He argues technology is our one greatest hope for securing and improving our future. Luckily for us, we have that technology: Artificial Intelligence.
AI is not just coming anymore — it’s already here. As ChatGPT has catapulted into ubiquity since its release to the general public in November of 2022, we’ve watched in awe as it’s transformed our landscape of possibilities. If current estimates coming out of Goldman Sachs are correct, AI could eventually raise annual global GDP by 7%, representing trillions of dollars of new economic activity. Generative AI alone could add between $2.6–4.4 trillion. How does this happen? There are three key ways through which generative AI may raise output potential: AI will simply be more productive, workers in occupations that are partially exposed to AI automation will apply some of their freed-up capacity toward alternative productive activities, and workers who are displaced by AI will become reemployed through upskilling.
There are plenty of historical parallels to support this theory. Following the internet boom, new occupations like UX designers, social media marketing professionals, and software developers emerged. These jobs not only raised aggregate income, but they also indirectly increased demand for service workers across different industries (i.e., retail, food service, and healthcare workers). Similarly, when Henry Ford rolled out the assembly line in 1913, he saved time and improved quality by allowing for increased specialization. But the new technology didn’t just help him — he expanded production and employed a massive labor force (52,000 people strong!) while famously paying employees $5 per day of work at a time when the going rate hovered around $2.25. We see that improving the technology in his factories shifted the aggregate output curve up — with more technology, more was produced. New technologies increase the quality of life, wages, and number of available opportunities over time.
Envisioning a future with AI
Fifty years ago, imagining a smartphone equipped with professional-grade cameras, access to all of the information on the internet, and the entirety of the contents of the yellow pages would have been a creative feat. Now, something like the contacts list isn’t just a standard feature — it’s a profoundly pedestrian logistical tool. This technology is now embedded within our everyday lives.
We’re starting to warm up to AI in the same way. ChatGPT saw a mind-boggling 100 million users just two months after its release (this milestone took nine months for TikTok to reach and two years for Instagram). The website is well on its way to verbification (“Let me ChatGPT that really quick”), joining the ranks of other pervasive technologies like Google and Photoshop. This is for good reason — it’s a useful tool! Generative AI can substantially increase labor productivity. McKinsey projects labor productivity growth of 0.1 to 0.6 percent annually through 2040, depending on the rate of technology adoption and redeployment of worker time into other activities. But ChatGPT and other AI plug-ins are just the tip of the iceberg. We’re moving towards co-pilot with generative AI, but the future lies with auto-pilot AI. Imagine a world where AI doesn’t just do what you need it to do — it actually anticipates what that need is for you. Appointments schedule themselves, you never run out of paper towels, and you come home after work to your favorite Thai food waiting at your doorstep. That’s a world I want to live in.
Obviously, before getting to this state of AI-driven bliss, we’ll have to navigate some serious ethical and practical dilemmas. Making sure that we build safety bumpers into our technology is of the utmost importance. We need to eliminate bias and error. We have a long way to go until we can make that leap of faith and reach auto-pilot AI. But, if we can trust it, that leap will unlock a whole new productivity frontier that could offset the coming economic collision of population and productivity decline.
The path ahead
Our world will be confronted with significant problems in the years to come. Fortunately, we have reasons for optimism: AI is here, and it gives us the power to change the curve of human history in very tangible ways. Of course, adopting AI won’t necessarily be easy. The launch of ChatGPT and ensuing public enthusiasm for conversational AI has been exciting, but it represents only the first novel step into a much more significant race to leverage AI for productivity enhancement. If we can use AI to make ourselves exponentially more efficient, we can sustain our growth and development. The real question now is whether we can adopt AI, and adapt with it, fast enough to offset the population challenges on our doorstep. Leading with trepidation won’t get us there, but leading with shared optimism and a collective assertion of ethical standards will.
