Bet 868
Duration 9 years (02021-02030)
PREDICTOR
Erik Brynjolfsson
CHALLENGER
Robert J Gordon
will go to Givewell if Brynjolfsson wins,
or Givewell if Gordon wins.
I acknowledge that productivity growth in the US has been disappointing for over a decade and the US Congressional Budget Office (CBO) predicts that it will average just 1.5% from 2021-2025 and 1.4% from 2026-2031.
However, I believe we are in the early stages of a productivity boom that will surpass the official predictions.
There are three reasons for my optimism:
1. AI is a general-purpose technology that is affecting almost every industry while accelerating the pace of discovery. Recent breakthroughs in machine learning will boost productivity in areas as diverse as biotech and medicine, energy technologies, retailing, finance, manufacturing and professional services.
2. The productivity benefits of general-purpose technologies typically take years to show up in the official statistics. In fact, productivity is initially suppressed as organizations invest time and effort creating intangible assets like new business processes, new skills, new goods and new services. However later, these investments are harvested, boosting productivity. The result is a productivity J-curve. Recent research indicates that are approaching the rising part of the productivity J-curve for the AI and related technologies.
3. There are several temporary factors which will boost productivity even further.
a. The Covid relief bill and associated spending, which is reducing the slack in the economy, creating greater incentives for businesses to implement productivity improvements.
b. the accelerated adoption of digital technologies due to the pandemic.
Further explanation for my optimism is described in my article with George Petropoulos “The Coming Productivity Boom”: https://www.technologyreview.com/2021/06/10/1026008/the-coming-productivity-boom/
Where does 1.8 percent fit into the historical record? For perspective consider the productivity growth record over five intervals: 3.2 percent per year (1948-72), 1.5 (1973-95), 3.3 (1996-2004), 2.1 (2005-2009), and 1.1 (2010-2019).
My interpretation is that there are substantive factors that explain the pattern of ups and downs. A useful precedent is the 23-year-long slowdown period that began in 1973. This reflected the exhaustion of the growth-enhancing benefits of epochal inventions of the late 19th century, particularly electricity and the internal combustion engine that had maintained growth at a rate of 3 percent all the way from 1920 to 1972.
The computer revolution that converted business practices from typewriters and file cabinets to flat screens, electronic storage and search engines, was powerful enough to revive growth to 3.3 percent, but for only nine years, not 50 years. As the digital conversion was completed around 2005, productivity growth slowed back to only 1.1 percent in 2010-19, the slowest in American history.
Will robots and AI bring a new revival comparable to the post-1995 digital revival? A reason for doubt is that a doubling of the U.S. stock of robots in the past decade failed to revive manufacturing productivity growth, which has languished at a pitiful 0.1 percent rate. Similarly, AI is nothing new, and for more than a decade has replaced human customer service representatives by annoying voice-recognition systems without reviving growth. Much economic activity, from home construction to slicing deli meats at the local supermarket, remains immune from a radical AI transformation.
Bet will be adjudicated using US Bureau of Labor Statistics when the 2029 Q4 data becomes available. If this source of data is not available, Long Now will use the closest analogous source of information.
