Ai Could Trigger The Biggest Productivity Boom Ever

I’m on a deadline that is approaching fast to finish writing my next book, The Great Progression: 2025 to 2050. Luckily, my use of artificial intelligence (AI) tools has dramatically increased my productivity, so I should hit that deadline. But the increase in my own productivity has gotten me thinking a lot more about what happens when all knowledge workers get a lot more productive with AI.

I have written two comparable books in the era before AI, and each time I had a human research assistant to help me. They transcribed my interviews and verbal notes, located material I needed to read, researched areas I did not have time to explore myself, and helped fact-check my writing.

My AI research assistants have made me at least twice as productive as I was when I wrote my previous books with human assistants. There are obvious examples of how — AI needs mere seconds to transcribe interviews that would take my human assistants all day, and it can pull up any fact or concept in context, across virtually any field, almost instantly.

But there are less-obvious productivity boosts, too. These arise in how I use Google’s NotebookLM. Since Amazon introduced e-books in 2007, I’ve read almost all my books on a Kindle. I highlight all the passages that make an impression on me using various colors to mean different things. For this book project, I exported all of those highlighted passages from Kindle and then imported them into NotebookLM.

Now, I can ask my Gemini assistant to pull from that text the six passages most relevant to any topic — for example, productivity growth rates — and it will provide an excellent list in 30 seconds. The AI has read everything of note that I have read in the last 20 years and can retrieve and explain it on demand — a human assistant could never do anything remotely comparable.

This gets even stranger and more amazing. A couple times a week for most of my career, I have recorded verbal notes about what I am thinking. It’s something like an audio journal. I talk through the ideas I am wrestling with or the problems I am trying to solve in a shorthand that only I can readily understand. These verbal notes become especially important as I rough out chapters and passages week by week for a new book I’m writing. With my previous books, my poor human research assistants had to try to transcribe the notes despite having essentially no idea what I was talking about.

This time, I had an AI transcribe 18 months’ worth of these verbal notes and then imported the transcripts into my NotebookLM. What then happened shocked me. Gemini essentially learned my personal shorthand language and could retrieve obscure thoughts that I had roughed out months ago. My wife of 35 years wouldn’t have been able to understand almost any of it. Yet the AI figured it out.

I can already see a through-line to becoming as much as 10 times as productive at writing a book within the next year or two. Generative AI has only been mainstream for several years and is still improving every month. I am only just getting used to these new tools and understanding how best to use them. I am incredibly augmented today and maybe two to three times as productive, but with teams of agents assisting me in the near future, I will probably reach 10x.

Soon, all knowledge workers — and eventually most workers — will experience their own big productivity boosts. AI is particularly well-suited to work in the rational world of coding, and so in 2025, software engineers were among the first categories of knowledge workers to be augmented and enhanced by it.

We did not see mass layoffs of software engineers last year, though we did see some layoffs and adjustments to hiring out of college. What we mostly saw was an increase in productivity throughout the profession. So far, the first companies hit have jumped on the productivity gains and mostly leaned into more growth rather than cost- cutting. Set aside worries about job disruption for now (we’ll come back to that), and stay with me as we talk through the implications of the rise of productivity rates throughout the whole economy.

Small increases in productivity growth rates make big differences over time

Economists are obsessed with productivity growth rates because they are arguably the best indicator of an economy’s long-term health. When the people of a society are consistently more productive, the overall economy grows and more wealth is generated. Wages then can rise, and the standard of living improves. The society gets steadily richer without workers having to work more hours. In some cases, they can even work fewer.

Economists track two numbers in the aggregate: labor productivity and a broader metric known as Total Factor Productivity (TFP). The first measures how much output you get for how much labor — for example, if one hour of work generates $100 in wealth one year, but $110 the next, labor productivity has increased. The second measures the efficiency with which an economy translates a bundle of inputs — labor, technology, capital, etc. — into outputs. If TFP increases by 2% from one year to the next, that means the economy can produce about 2% more output than the previous year using the same inputs. In short, it has become more efficient.  

TFP is the more important of the two metrics and the one I’ll use going forward. For simplicity, I’ll just call it “productivity.”

Small differences in productivity can result in hugely different outcomes over time. Even if it increased at a rate of just 2% each year, that 2% would keep compounding off a growing base, much like a savings account. At the end of a generation, you’d have a lot more than what you started out with.

During the post-World War II economic boom — the 25 years from 1948 to 1973 — productivity in the United States increased at a rate of 1.9% annually. That meant that the average worker could produce 60% more at the end of that period than at the beginning, creating more wealth that could be shared throughout the society. We often talk about that period as a Golden Age where all boats rose with the tide.

In more recent times, the U.S. has been experiencing smaller annual increases in productivity. In the aftermath of the Great Recession — the dozen years from 2007 to 2019 — it increased by just 0.6% annually. If you took that rate and compounded it over 25 years, you would end up with the average worker only producing 16% more — compared to the 60% more from that postwar era. That’s a lot less economic growth, fewer wage gains, and a much more stagnant standard of living. We remember those times as tough, with many people dissatisfied.

We know that new general-purpose technologies bring productivity gains, though it’s not a direct or immediate correlation. Businesses have to adopt the new technologies and rework their processes, and workers have to retrain and adapt to take advantage of the new tech’s potential.

This did happen during the internet boom of the late 1990s and early 2000s — between 1995 and 2007, productivity increased by 1.4% annually. Some of this increase was due to factors beyond just the internet, such as businesses fully adopting personal computers and other parts of the digital transformation.

If you took that 1.4% annual increase and compounded it over 25 years, the average worker would be producing 42% more by the end of the period — significantly more than the 16% growth during those slow post-Great Recession years — and we generally remember wage gains for most workers and a rising standard of living in that internet boom time.

What happens if AI boosts productivity growth rates above historical norms?

Now let’s look ahead at how the arrival of AI, a new general-purpose technology, could affect productivity. I talked about how I could potentially be 10 times more productive in writing books in the next year or two, but the productivity of the whole economy is highly unlikely to increase by that much (though there are some extreme techno-optimists who claim otherwise).

But what if the widespread adoption of AI could lift overall productivity by something like 2.5% a year compared to the 2% a year during the postwar boom? That rate increase would compound over 25 years to an 85% gain compared to the 60% gain of that Golden Age. 

Or what if AI pushed productivity to increase by 4% annually — twice the rate experienced during the postwar boom? This rate is historically unusual but not unprecedented — we’ve seen it in China and in Europe and Japan during the rebuilding phase that followed the devastation of World War II.

If you could keep that 4% increase up for 25 years, you would arrive in 2050 with a 167% gain in what one average worker could produce. If a company was paying a worker $100,000 a year in 2025, they could afford to pay them the equivalent of $267,000 in today’s dollars in 2050.

If you assume that the size of the labor force in a developed economy like the U.S. grows modestly over that time (say, about 0.5% annually), the country would be experiencing 4.5% GDP growth per year, compared to the roughly 2% GDP growth per year in recent decades.

There would be a lot of negotiation around what to do with all that new wealth, but the tax base could explode — we could eliminate federal budget deficits, fund more social programs, or reduce inequality.

That scenario — a 4% annual increase in productivity — may sound crazy at first. We could increase productivity at twice the rate the U.S. pulled off after World War II? But then again, those Americans did not have intelligent machines and autonomous robots to help them.

Some AI experts think productivity could increase by much more and much faster than what I’ve laid out. That group includes the people who talk about AI achieving super-intelligence and freeing itself from human control. I don’t believe we will hit that between 2025 and 2050, the 25 years I’m calling the Great Progression. Increases in productivity depend on human beings adopting and adapting new technologies — and that will take some time.

The new rebalancing act of wealth coming off AI

AI is going to boost productivity and drive economic growth. This is going to happen under almost any scenario going forward. An enormous amount of new wealth is going to be created in the next 25 years, and it is going to significantly disrupt the way the economy works.

If we distribute the new wealth in the U.S. the same way we have been for the last 40 years, the billionaires and 1% elite who already own a lion’s share of the wealth are going to pull even farther ahead of everyone else, and the government will remain underfunded and overburdened. The many workers who will have their careers and livelihoods disrupted by AI are going to be even more angry and scared. This is a recipe for a big backlash or even efforts to tear down the existing system.

On the other hand, we could use the windfall of new wealth coming from AI to fundamentally rework the economic system to work better for everyone over the long haul. There will be plenty of wealth to go around as we move toward a society of abundance. Many leaders in the tech world understand the value of giving all employees equity in their company, and so they can understand the logic of giving everyone in society a stake in the AI economy writ large. We have an opportunity to create a new economy based on a transformative new technology. How should that new economy operate?

A growing number of people think one fundamental rework of the economic system should be going from redistribution at the end of wealth creation to predistribution of wealth itself. Everyone needs a stake on the capital side of the AI revolution. Not necessarily an equal stake, but a stake. Everyone needs a piece of equity in the AI economy.

Over the last 80 years, the federal government used taxpayer dollars to fund much of the research that laid the groundwork for today’s manifestation of AI. AI companies draw off their employees, who were often educated in university systems we helped fund, and who drive to work each day on roads built and maintained using our tax dollars. For that matter, the large language models supporting today’s most powerful AIs were trained on the internet — content that we created collectively.

The wealth to spin off this technology shouldn’t go solely to founders, VCs, and a handful of tech employees. They certainly deserve a good chunk of it as a reward for all their creativity and hard work. But we need to rethink who gets what share of the new wealth that’s coming in the AI Age.

The new notion of Universal Basic Capital for all

This is where the notion of stakeholder capitalism — a concept that has been talked about for a long time — gets real. How do we move from the longstanding status quo of financial capitalism — where those with money can keep compounding their wealth — to a new form of capitalism where all the stakeholders — workers, local communities, and governments of various levels — partake in it? How can everyone benefit from the growing value of stock markets or the economy at large?

Universal Basic Capital (UBC) is one possible answer. It generally means giving every citizen a share of productive capital so they can earn capital income in the long run. In the AI context, this would mean figuring out how to make sure everyone in the U.S. gets to tag along as the companies riding the AI revolution grow in value. This could mean that a certain portion of stock or a meaningful percentage tax gets pulled from AI companies — these could be the trillion-dollar Big Tech companies building the AI models, any AI company hitting a certain billion-dollar valuation, or perhaps all companies in AI who grow beyond some minimal threshold. That wealth could take the form of a stake that goes directly into the federal government, a new American sovereign wealth fund, or even the accounts of individual citizens.

People like OpenAI CEO Sam Altman and other serious tech leaders are talking about how something like UBC might work and how their industry might fund it. They understand how much wealth AI will generate and how much disruption to the current economy might lie ahead. They are not out to squeeze all the juice out of AI for themselves. They are open to figuring out win-win solutions that keep all Americans moving ahead. Their opening bids on what they are willing to contribute might be too minimal to make a big difference. And they might want their companies to shoulder more of the burden of funding the public pool, rather than contributing to it from their personal wealth. An ongoing negotiation in the politics of progress is playing out, and I predict the people (via the government) will be able to work something out with the AI industry and those with wealth.

To be sure, there are those in tech who want the existing system to stay locked in place — they want to remain the main beneficiaries in this next phase of wealth creation. Even modest talk of new forms of wealth taxes on billionaires in California has some of them howling and threatening to move to states with no chance of implementing such taxes, like Texas.

Rebalancing the inequalities of wealth that have built up in our current Gilded Age is going to be difficult. It was extremely difficult coming off the original Gilded Age, which I have laid out in a previous essay. The Robber Barons and those with great wealth in the early 20th century fought every reform at every stage. But we reset the system then, and we will do so again now.

A rare opportunity to reset the economy for a world of abundance

At this juncture, America needs to really rethink its economic fundamentals, just like it did during the three previous reinventions I laid out in other essays. How much wealth should any one individual deserve to hold before they have to contribute a much higher percentage to the shared pot? How much wealth is too much? Who really needs more than $1 billion? Do most of us really want to live in a democracy where all citizens are supposedly equal yet a handful of billionaires can spend millions of dollars to impose their will on elections in ways that the rest of us can’t? Should we restrict how much wealth can be used to influence politics? How much of that accumulated wealth can they pass on to their children? What is a reasonable inheritance tax to ensure we don’t create a permanent class of elites, a plutocracy?

These are reasonable questions for Americans to revisit at these reinvention junctures, when we reset our systems to work much better going forward. In the last such juncture, the one coming off World War II, the answers to those questions were very different from what we consider normal today.

Those with great wealth at the top of postwar society — the billionaire equivalents — were taxed at 90% on income over $400,000 (equivalent to $4.8 million today). That rate held for 20 years — through most of the great economic boom — until it was reduced to 70% on income over $200,000 (equivalent to $2 million today) in 1965, and which held until 1976. The range has changed several times since then, and today, the top federal marginal income tax rate is 37% on income above $609,350.

The American people, through democratic politics, using their elected governments, have fundamentally reworked the rules by which the minority of those with great wealth play before. We now need to do it again. My forthcoming book will go into more depth on how.

This article AI could trigger the biggest productivity boom ever is featured on Big Think.