TED: “The key to growth? Race with the machines”

Erik Brynojolfsson, director of the MIT Center for Digital Business, gave a great TED talk on the recent changes in world-wide productivity, computers, machine learning and the “great stagnation“.  There has been a stagnation in GDP per capita for the last 5 years in the United States, Europe, and Japan.  However, growth has continued unabated in India, China, and Brazil.  Economist Tyler Cowen is not particularly optimistic about the current economic situation which he describes in his book “The Great Stagnation: How America Ate All The Low-Hanging Fruit of Modern History, Got Sick, and Will (Eventually) Get Better” ($4 Kindle edition). Erik disagrees with the idea that the economy, technology, and science are stagnant.  Instead, we are being confused by what he calls the “great decoupling” of productivity growth and employment/median income.  He believes that we are actually in the early phases of the “new machine age” which will create more wealth than the industrial revolution.  Here are some notes on his TED talk:

  • The introduction of electricity did not immediately lead to large productivity growth. At first, factory managers just replaced existing steam machines with electric machines. It took thirty years (one generation) before the factories were redesigned with electricity in mind and new work processes were created. Then productivity growth took off.
  • Big general purpose technologies like fire, agriculture, writing, metallurgy, printing press, steam engine, telecommunications & electricity, and now computers cause “cascades of complementary innovations”.
  • The GDP growth rate in America has been fairly steady at an average of 1.9% per year since 1800.
  • Erik graphically compares the 1960-2011 productivity growth caused partly by computers and the 1890-1940 productivity growth caused by electricity.  They appear quite similar.
  • Productivity has grown faster (2.4% per year) in the last decade than any other decade since 1970. The average productivity growth rate was about 2.3% during the second industrial revolution.
  • World GDP in the last decade has grown faster than any other decade in history.
  • New growth has been driven by thoughts and ideas more than physical products. This growth is hard to measure. The massive utility of free products like Google Search or the Wikipedia is not included in GDP.
  • The “new machine age” is “digital, exponential, and combinatorial”.
  • Digital goods are perfectly, freely replicable and they have near zero transportation cost (often moving at great speed crossing the globe in a minute.)
  • Computer power has grown exponentially. In Erik’s words, “Computers get better faster than anything else ever.” (2013 playstation = 1996 supercomputer)
  • We are not designed by evolution to anticipate exponential trends. We expect linear trends.
  • We now have Big Data and Machine Learning which allow us to analyze everything more deeply.
  • He disagrees with the idea of “low hanging fruit” inventions (see e.g. Cowen’s book). He thinks that “each innovation produces building blocks for even more innovations”.
  • The digital, exponential, and combinatorial nature of the new machine age will lead to the greatest industrial revolution in history. (see e.g. cat transportation at 6:55 in the video).
  • Machine Learning may be the most important innovation of this revolution. (The Watson Jeopardy team improved faster than any human could by using machine learning.)  Watson’s technology is now being applied in the legal, banking, and medical fields.
  • “The great decoupling” is the decoupling of high productivity growth from employment and median income.
  • Jobs are being replaced by computers.
  • In 1997 Gary Kasparov was beaten by the super computer Deep Blue. Today a cell phone can beat a chess grandmaster.
  • In computer assisted chess play (Advanced Chess), sometimes the winners are neither computer experts nor grandmasters. Instead, they were the best at interacting with the computer—guiding its exploration of the possibilities. So, maybe we will not lose our jobs to computers, rather we will collaborate with computers to create wonderful things.


  1. hundalhh’s avatar

    One of my friends (Mary H) emailed this response:

    What is the purpose of machine learning, Watson, and “growth” if machines do it and we don’t? What is the purpose of living and how does this apply to it? We’re already destroying Nature by fracking shale for natural gas, how much worse does it have to get before people realize that “growth” moves us further faster from where we need to evolve? If I were to imagine myself taking a train ride, I imagine looking out the window and enjoying the scenery (cows, factories, small towns, busy centers, a patchwork quilt). When I think of traveling on a bullet train, I imagine myself on a handheld device and the outside landscape is irrelevant because the train is traveling too fast to take it in. If we are developing technology faster than we can take it in, what will life be like for our grandchildren? Will being smart be an advantage? Or will it be all about having the quickest internet (or whatever it might be called then) feed to come up with the info faster than peers? It becomes a competition with a different skill set; one that is more about salesmanship of ideas, knowledge, and “what’s considered true right now” than solidity in what is traditionally considered true, i.e. I know this because I study this field, rather than, I know this because I looked it up on the internet, and so it must be true.

  2. hundalhh’s avatar

    John F emailed this response:

    It is true that a vast amount of very valuable stuff is now almost or totally free. Great works which five years ago were expensive in print, such as McCauleys works and Parkmans, are now free on my Kindle. To some extent this will put people out of work, especially the scribbling classes. Almost no history I know correctly portrays the Industrial Revolution in England. They think it created a vast number of poverty-stricken people. It didn’t. The cause was the agricultural revolution which began maybe about 1750. See Goldsmith’s “The deserted village, 1763??”. He describes a rural village depopulated by changes in farming methods. This is the source of all those poverty-stricken people. The factories gave some of them work. Others came to America, others starved.

    Think how many musicians were put out of work by the phonograph, tape recorder, etc.

    You should also read a piece Ben Franklin wrote on population in 1751 – 13 pages. Adam Smith kept two copies on his desk. Malthus one.

    There was process used by landowners to dispossess farmers – it was called enclosure. A rich land owner dispossessed his tenants and turned his land over to sheep, which was more profitable. Everybody knew about this then.

    In America the conditions were different: my uncle had an old pamphlet about how to farm, a self-help book for farmers (very American). It described all the tasks such as planting, weeding, and reaping. Most required two men and a boy. It was a great help to have a wife who could produce boys. The book Albion’s Seed points out that if a New Englander and his wife both lived to be fifty they averaged ten children. This is why the population doubled every twenty years! But that was all right because they could just go west and buy another farm.

    In England at this time the birthrate was lower, but there was no new land.

    (Population in South Carolina did not increase this way. More people died from diseases than were born and new people had to come to take their place (until about 1772).)

Comments are now closed.