π€π America's AI Boom: an update
New US productivity data are what you would expect if the US economy were about to accelerate. Fingers crossed.
Quote of the Issue
βIndividualism stimulates innovation by not penalizing heterodox intellectuals who come up with unconventional and possibly heretical ideas and think outside the box.β Joel Mokyr, A Culture of Growth: The Origins of the Modern Economy
I have a book out: The Conservative Futurist: How To Create the Sci-Fi World We Were Promised is currently available pretty much everywhere. Iβm very excited about it! Letβs gooooo! β©πβ‴π
The Essay
Hereβs a scary counterfactual: What if β¦ in November of last ye ar, a prominent Chinese technology company (think Baidu, Tencent, or Alibaba) had unveiled a generative AI system far more advanced than what any American company was developing? This shocking event would have been likely seen as threatening a significant economic and military edge for China, potentially reshaping the global balance of soft and hard power. Both politicians and the media would have treated such a Chinese AI breakthrough as something akin to Sputnik in the 1950s, or even worse. Maybe more like if the US had lost the Space Race to the Moon or the Soviet Union had exploded the first atom bomb.Β
In this speculative, alt-history scenario, I would guess that Taylor Swift wouldnβt be Time magazineβs Person of the Year for 2023. Instead, the honor might have been awarded to the leader of a hypothetical US initiative, maybe named something like "Project Prometheus." This project, possibly led by tech figures such as Sam Altman, Elon Musk, or Marc Andreessen, would represent a major national effort to ensure the United States achieved artificial general intelligence, or AGI, technology before China.
Yet even in our own reality, a few years hence, we might look back on Timeβs Swift selection as an odd one. Another bit of speculation for you: What if 2023 marks the start of an AI-driven productivity boom at least as strong as the tech boom of the late 1990s and early 2000s β if not even more powerful and longer lasting? 2023 might then be viewed as a great inflection point marking the end of what I call the Great Downshift in my book and what some others term the Great/Long Stagnation.Β
If youβre looking for a bit of suggestive evidence today of that sunny techno-optimist tomorrow, here you go, courtesy of a new analysis by JPMorgan economist Daniel Silver:
Nonfarm productivity growth was revised up from 4.7% to 5.2% saar in 3Q, beating expectations somewhat and making an already strong quarter for productivity look even firmer. The data can be noisy, and we don't think the 3Q reading in isolation is the underlying trend (just like we donβt think the strong GDP growth reported for 3Q should be viewed as the trend). But there has been some recent firming in the productivity data more broadly, with the updated figures showing a 2.4% oya gain into the third quarter.Β
Caveat: Productivity numbers are notoriously volatile and subject to revision. As my AEI colleague Michael Strain puts it: βI think it's never a good idea to ignore new information. And I think my assessment of whether or not we are at the beginning of a period of increased productivity growth is brighter today than it was yesterday as consequence of these new data. [Yet] we have not seen quarters and quarters and quarters of stronger productivity numbers.β (Keep in mind that worker productivity growth is the key driver of higher incomes and living standards over the long run, especially if driven by rapid innovation. Faster productivity growth to support fast wage growth would be especially welcome right now in helping the Federal Reserve hit its inflation goal without more interest rate hikes β or even get to rate cuts more quickly and thus lessen recession risks.)
Point taken. Still, the 3Q number comes after a strong 2Q of 3.6 percent, making for one of the strongest six-month periods in some time. And more than one economist thinks advances in AI, including progress in machine learning that predates todayβs amazing GenAI chatbots, may have already set the stage for faster productivity and economic growth in the years ahead.
Iβve already written about cheery AI forecasts from Goldman Sachs, so here are RSM economists Joseph Brusuelas and Tuan Nguyen, whose cheery baseline 2024 forecast looks for 1.8 percent real GDP growth and inflation below 3 percent (bold by me):
One encouraging development this year has been rising productivity, which increased by 4.7% in the third quarter [now 5.2%] and averaged 4% during the previous six months. With growth in the third quarter advancing at the robust 2.9% rate on a year-ago basis even as hiring cooled, productivity gains, and not necessarily labor, have been fueling these gains. Should this increase in productivity continue, our 2024 economic outlook may ultimately be somewhat restrained and the economy could increase its pace of output without stoking inflation fears.
It is clear that firms are continuing to make investments in software and intellectual property to offset growing labor scarcity. Over the past decade, productivity increases have averaged 1.5%, which is soft compared to the 2.3% of the previous 10 years and the robust 2.8% during the 1990s productivity boom . β¦
In one of our alternatives to the baseline, the economy will prove to be far more resilient than expected, with growth approaching 2.5%. The combination of excess savings, which we have estimated to be as high as $1.3 trillion, strong job demand and rising real wages amid soaring productivity will propel an unexpectedly strong U.S. economy. Under this scenario, the major surprise will be the impact of large firmsβ integrating sophisticated technology, most notably artificial intelligence, which will turbocharge output amid restrained labor costs.
Or as economist Erik BrynjolfssonΒ recently toldΒ Fortune:
I see a coming productivity boom. I see about a doubling of productivity growth in the coming decade as a result of these technologies. Itβs not as simple as just buying the software or hardware. General purpose technologies like electricity, the steam engine, early computers, often take a decade or more to translate into significant productivity gains because of all the changes you have to make in business processes and rescaling. This time itβs definitely happening faster. β¦ For a lot of the problems we haveβthe federal budget, health care, these big, nagging problemsβproductivity is like an elixir that makes all those problems much more solvable.
Or this from Capital Economics in the UK:
Generative artificial intelligence (AI) would appear to have all the characteristics of a βgeneral-purpose technologyβ that will revolutionise economies and has the potential to deliver a substantial boost to productivity growth. β¦ Our proprietary AI Economic Impact Index, which ranks countries according to their ability to innovate, adopt and adapt to AI, suggests that the US will lead the AI revolution. Along with beneficiaries including Singapore and South Korea, US productivity growth will accelerate to over 2% a year in the 2030s. β¦ One consequence of the AI revolution is therefore likely to be greater US economic outperformance of Europe. β¦ China will lead some aspects of the AI revolution but will struggle in others. This is in part because AI is likely to become a new fault line in global economic fracturing, which will reduce Chinaβs access to US technology and mean it will need to develop AI capabilities domestically. Moreover, while we expect AI to boost productivity growth in China, this will be more than offset by structural headwinds pushing in the other direction. A combination of weaker growth in China and the boost to US productivity from AI means that we expect the US to remain the worldβs largest economy over the coming decades.
My message to Washington: Letβs not screw this upΒ through heavy-handed and preemptive regulationΒ based on science fictional fears or on wanting some ill-advised βdo overβ on internet regulation. Faster, please!
Micro Reads
βΆ Itβs Time to Fall in Love with Nuclear Fusion β AgainΒ -Β Virginia Heffernan, WIRED |
βΆ Google DeepMindβs Demis Hassabis Says Gemini is a New Breed of AI - Will Knight, WIRED |
βΆ NASA says SpaceXβs next Starship flight could test refueling tech - Stephen Clark, Ars Technica |
βΆ Google Announces AI System Gemini After Turmoil at Rival OpenAI - Miles Kruppa, WSJ |
βΆ Boosting Laser Power Will Help Them Blast Drones - Charles Q. Choi, IEEE Spectrum |
βΆ Googleβs βGeminiβ makes mobile breakthrough for generative AI - Financial Times |
βΆ A Cultural Divide Over AI Forms in Silicon Valley - Ellen Huet, Bloomberg |
βΆ Elon Muskβs AI Dreams Are Going in Circles - Parmy Olson, Bloomberg |
βΆ How Nations Are Losing a Global Race to Tackle A.I.βs Harms - Adam Satariano and Cecilia Kang, NYT |
βΆ Google Updates Bard Chatbot With βGeminiβ A.I. as It Chases Chat GPT - Cade Metz and Nico Grant, NYT |
βΆ Accelerating science through evolvable institutions - Jason Crawford, The Roots of Progress |
βΆ India reveals that it has returned lunar spacecraft to Earth orbit - Eric Berger, Ars Technica |
βΆ IBM releases first-ever 1,000-qubit quantum chip - Davide Castelvecchi, nature |
βΆ Why We Should Settle Mars - Robert Zubrin, Quilette |
βΆ IBMβs Jerry Chow on the future of quantum computing - Nilay Patel, The Verge |
βΆ Should Government Regulate AI Technology? Readers Weigh In - Demetria Gallegos, WSJ |
βΆ Inside the A.I. Arms Race That Changed Silicon Valley Forever - Karen Weise, Cade Metz, Nico Grant, and Mike Isaac, NYT |
βΆ A US Nuclear Revival β and Net Zero β Depends on Westinghouse - Jonathan Ford, Bloomberg |
βΆ Removing zombie-like cells may hello treat multiple sclerosis - Grace Wade, NewScientist |
βΆ McKinsey Sees AI Adding Up to $340 Billion to Wall Street Profit - Aisha S Gani, Bloomberg |
βΆ The true extent of damage to schools from Covid-19 - Amy Borrett, Financial Times |