Everything from magazines to medical records is migrating to the digital world. In what ways is all of this affecting economics and society?
Brynjolfsson: Digital technologies have very different economic characteristics than technologies of the past, such as those that drove the Industrial Revolution. For instance, digital goods can be copied and those copies can be transmitted anywhere on earth at the speed of light at essentially no cost. What's more, the copies are perfect replicas of the original. Digital technologies can create enormous value. But they can also reduce the need for certain kinds of jobs, thus destabilizing entire areas of work.
Is this process accelerating?
Brynjolfsson: Yes, at an exponential rate. Ray Kurzweil, an inventor and entrepreneur, compared this process to what happens if you put a single grain of rice on the first square of a chessboard, put two grains on the next square, and keep doubling the amount with each square. By the time you get to the 64th square you have a pile of rice that's bigger than Mt. Everest. And yet, if you look at just the first 32 squares, the amounts of rice are quite manageable, amounting to only a few truckloads by square 32. Our society is now starting to enter the second half of the chessboard. The scope of digital technologies is expanding because of the combinatorial nature of innovation. In a combinatorial economy, each idea, rather than using up the stock of ideas, creates building blocks for other ideas. And the new combinations of ideas can be even more valuable than the original ones.
What are the economic implications of all of this?
Brynjolfsson: I believe that we will soon be entering a period of accelerating economic growth. However, we will need new tools to measure what is happening. For instance, software is poorly measured in the U.S. GDP and services that are free are not measured at all. Yet look at Facebook, YouTube, Wikipedia — these are all free services. So as people use Wikipedia, for instance, instead of buying encyclopedias, they have access to knowledge without paying for it. But in terms of official GDP, it is a big zero. And if you look at official statistics, the share of the information economy as a percentage of GDP hasn't budged in years. Much of the information economy is invisible because the tools for measuring GDP were invented long before the information economy became important.
What are the consequences of digitization for the job market?
Brynjolfsson: Manufacturing employment, for instance, is declining across the board. Today, there are about 20 million fewer people working in manufacturing in China than there were in the late 1990s. That's about a 20 percent decrease. Why? A robot can perform a simple job such as packing boxes for about $4.00 per hour; but it can work 24 hours per day. So if you have a high school education in the U.S. or are a Chinese factory worker, you are in the bull's eye. It is not that jobs are migrating from the U.S. to China, but that manufacturing jobs in both countries are migrating to robots. The result is that in the U.S. today, although GDP has grown significantly and the net wealth of the nation is at a record $77 trillion, the median income is no higher than it was in the late 1990s. Instead, wealth has been concentrated in a relatively small share of the population. Unfortunately there is no economic law that says that technological progress has to benefit a majority of people.
Are white collar jobs also at risk?
Brynjolfsson: Yes. Take jobs like accounting and bookkeeping, for instance. Today there are around 17,000 fewer tax preparers in the U.S. than there were only a few years ago. They have been displaced by software. What's more, the companies that produce such software can be run with very few people. For instance, Instagram, which specializes in sharing photographs, was sold for over $1 billion yet had a team of only 14 people. Compare that with Kodak, which had nearly 150,000 employees in the 1990s. The underlying trend is that when you shift from manipulating atoms to manipulating bits you need far fewer resources and have much more profit.
Would you call this good news or bad news?
Brynjolfsson: It should be good news because all sorts of information that's extremely expensive today — medical, legal, financial, to name a few key areas — will be much cheaper in the future, and overall income will be higher. But this is also very disruptive for those who are affected.
In other words, you expect digital technologies to benefit average people?
Brynjolfsson: Don't get me wrong. I think that there are huge challenges, such as those we've been discussing with regard to employment, but on balance what we are talking about is mostly good news. Take health care, for instance. Big data will increasingly be used to find patterns in drug interactions and in determining which treatments are most effective. And I believe that bioinformatics, primarily with regard to genomics, will be one of the big stories of the 21st Century. Average people will also benefit from improvements in urban infrastructures, including buildings, communications and transportation networks. The reason? Trillions of Internet-connected devices will collaborate in real time to optimize efficiency, thereby improving services while probably reducing costs.
And what about the energy sector?
Brynjolfsson: We are certainly making progress in terms of cheaper energy sources. And there are now steps in the direction of making electrical systems that are capable of communicating with information networks in order to combine their demand with renewable energy production. An extremely important story here is that we are witnessing an exponential improvement in the energy efficiency of computation itself. This is what is sometimes called “Koomey's Law.” Dr. Jon Koomey noted that the cost of computation — in energy terms — is improving even more rapidly than the technologies covered by Moore's Law. When you put all of this together — supply side improvements, remarkable advances in potential for demand management, and the improvement in the energy efficiency of computation — it makes me pretty optimistic about our energy future.
Where do you see the most rapid digitization?
Brynjolfsson: The leading areas today are the media, retailing, finance, and manufacturing. The ones that are coming are health care and education. These are two very positive areas. Take massive open online courses — MOOCs, for instance. These hybrid teaching systems have the potential of taking the very best lecturers and learning systems and replicating them so that they are available to millions of people. Furthermore, in such digital environments every interaction is measured and quantified, which leads to rapid learning about which techniques work and which ones don't. These characteristics can lead to a higher rate of improvement in learning and teaching.
Do you believe we are headed for technological singularity — the concept of machine intelligence eventually equaling or exceeding human intelligence?
Brynjolfsson: I think there's a lot of good science behind this idea. The engineers I know are confident that exponential improvements in artificial intelligence — AI — will continue to be realized for at least another decade. Eventually, it's likely that we will have computers that operate at human or superhuman levels of intelligence. We could ultimately have computers that design even better computers, and AI that creates even better AI. This is what people talk about with regard to singularity: a positive feedback loop that leads to very rapid acceleration, possibly within our lifetimes.
Could all of this lead to a world of non-interpretable results — systems that propose solutions that are based on such huge data sets that we cannot fathom them?
Brynjolfsson: This is an unfortunate side effect of many neural nets and big data algorithms. The results are accurate, but the explanations cannot be unpackaged for human consumption.
What should society be doing to prepare for an increasingly digital world?
Brynjolfsson: As Andrew McAfee and I say in our latest book, The Second Machine Age, technology is not our destiny; we shape our destiny. What that means is that we must be much more conscious about what kind of society we want to live in, what our values are, what our goals are, and explicitly demand this from our political leaders and ourselves. The technologies we've been discussing are enormously flexible and can be used to create enormous wealth. But that wealth can be concentrated in the hands of a small number of people or it can create widespread opportunity for billions of people. It can be used to create unimaginably destructive weapons, or to improve health care and to eliminate hunger. It can be used to eliminate privacy, or it can be used to enhance it. We realize that the failure to keep up with technology is hurting a lot of people right now. But if we do things right, it can help everyone. We certainly have it within our power to create shared prosperity. But that will never happen if we just close our eyes and assume it's going to happen on its own.