Times Square — a new series on jobs, automation and anxiety from the world's public square.
Brynjolfsson is part of the White House AI Select Committee announced at the US government’s AI summit earlier this year. His research examines the effects of information technologies on business strategy, productivity and performance, digital commerce, and intangible assets. At MIT, he teaches courses on the Economics of Information and the Analytics Lab.
Brynjolfsson was among the first researchers to measure the productivity contributions of IT and the complementary role of organisational capital and other intangibles. His research also provided the first quantification of the value of online product variety, often known as the “long tail”, and developed pricing and bundling models for information goods. His research has appeared in leading economics, management, and science journals and has been recognised with ten Best Paper awards and five patents.
Most recently, Brynjolfsson has co-authored Machine Platform Crowd: Harnessing our Digital Future (2017) and NY Times bestseller The Second Machine Age: Work, Progress and Prosperity in a Time of Brilliant Technologies (2014). His work on the limits and possibilities of machine learning have been published in the world’s leading publications, including Science.
Brynjolfsson and his fellow researchers are working at the frontier of our understanding on the inescapable trinity of the modern digital revolution: the rebalancing of minds and machines, products and platforms and of the core and the crowd. The uprising of the machines is stark when it changes cost structures; it’s stealth is equally potent when it creeps under the assumptions of incumbent companies whose processes are most resistant to change.
Equally, the paradox of unprecedented advances in machine learning and slowing GDP growth in advanced economies is a dominant theme in Brynjolfsson’s work.
“Depending on how they are used, machines, platforms and the crowd can have very different effects. They can concentrate power and wealth or distribute decision making and prosperity. They can increase privacy, enhance openness or even do both at the same time. They can create a workplace imbued with inspiration and purpose or one that is driven by greed and fear. As the power of our technologies grows, so do our future possi< style="color: #000000">bilities. This potential increases the importance in having clarity in our goals and thinking more deeply about our values”, write Brynjolfsson and his colleague Andrew McAfee in their latest book Machine, Platform, Crowd.
< style="color: #000000">Nikhila Natarajan spoke with Brynjolfsson on many of these themes during a New York-Cambridge, Massachusetts video call before the MIT Sloan professor leaves on a sabbatical to Stanford University.
< style="color: #000000">Excerpts from the conversation
covering the broad brush strokes of Brynjolfsson’s most acclaimed work on the confluence of machine learning, the future of work and firm level processes are captured below.
On blending tech and soft skills to complement the power of machine learning.
The people I see who are most successful are not necessarily the ones with math smarts or technical skills, but also a combination of the people and softer skills. Machines are getting better and better at doing structured tasks but machines are not very good at two categories where humans excel — creativity and large scale problem identification and being able to ask the right big questions and figuring out creative, out-of-the-box solutions. The second big category is interpersonal skills. Motivating, leading, selling, persuading caring for people and nurturing people — these are things that we don’t rely on machines to do. If you can combine technical skills with both creativity and interpersonal skills, that’s a winning combination.
On why firms continue to do “bizarre and counterproductive” things and how their processes blind them to the reality of the digital economy.
It’s very hard for old firms to recognise that they need to change. A lot of startups are dominating new industries. When some people at a firm do recognise that their processes need to change, it takes a coordinated effort for all the components to make the switch. The analogy I sometimes use is this.. let’s say you have an analog watch — maybe it’s a Swiss watch and you want to take advantage of it piece by piece. You open up the back of the watch and replace one of the transistors.. you know, it never works like that. You have to make a co-ordinated change. You have to have a completely digital watch or an analog watch. Bits and pieces don’t work. It’s the same with firms, it’s just not so obvious. With the watch, you can see the pieces and in organisations, it’s their incentive systems, their processes, hiring practices and each of them has to fit with the other practices and often no one person knows how it always fits together. Some firms, however, do succeed.
On the risks for firms that don’t reinvent their business processes early enough.
The risk is that you get stuck halfway. You have parts that are digitised and some that are not. Not only do you not get the full benefit but in some ways you can be worse off than if you hadn’t made those halfway measures. Sometimes there’s an element of luck in whether you are able to make the right choices and the invisible hand of the marketplace in economies where we see that transition, where we see many firms try. Some of them hit on the right combination and succeed. A firm that takes an entrepreneurial approach, a market driven approach, that is ready to experiment and test is more likely to succeed. Rather than having a 5 year plan, having a number of small experiments can be a more effective way to make that transition.
On the path for online businesses that are free, have no paywalls and rely almost completely on turbocharging paid searches to drive traffic.
A whole spectrum of outcomes.. what we’re seeing is that many of these content markets are looking like they have a long tail — there are a few winners and some medium firms and many small firms filling in niche markets. Look at the videos on YouTube — there are some that have a billion views and many many that have just a few 100 views. That’s very different from what you see in many other industries. Manufacturers don’t have the same kind of long tail but in digital markets, there are many more of these long tails. Many of them can be sustained by advertising especially as advertising becomes more targeted but that doesn’t necessarily mean you’re going to be in that billion views end of the spectrum. You see the same pattern in content markets as they become digitised.
On how/why success doesn’t (always) accrue to those with the best products or the best understanding of strategy.
It doesn’t always accrue to them. Certainly it always helps to have the best products and the best strategy. But it’s also important to understand that there are always some small circumstances that could tip the playing field one way or the other. For instance, let’s take the case of social networking. It makes sense for us all to be on the same social network so I can share pictures with my brother, my mother, my high school friends and if we were on different networks, we couldn’t do that as easily so there tends to be one dominant firm there because of powerful network effects. In the early 2000s, there were multiple social networks. Hundreds or even thousands of people had ideas for social networks but ultimately the economics are such that only one would dominate. That could be a matter of unusual circumstances, maybe some small effects that tilted the balance one way or the other, a key group of influential people or a slightly better user interface could lead Facebook to overtake MySpace.
On common mistakes people/businesses are making when they are up against platforms driving prices down.
The thing to understand is that when platforms drive prices down, that can be beneficial for all participants if they are part of that two sided network. The economics of two sided networks is very different from traditional networks. Having a large number of people participating in two sided networks driving prices down grows the market quite rapidly. It’s not the usual downward sloping demand curve. In a way, it’s turbocharged. It only helps if you are part of that two sided network, it doesn’t help if you’re competing against it. One of the common misconceptions is to fight the platform head-to-head rather than being part of it.
On the US government’s artificial intelligence strategy.
I was at the White House for that meeting. The Obama government had three very good reports
on this which went into great depth and and now the Trump administration is picking up on some of that. It is a very market driven approach which expects private sector to take the lead. Fortunately in the United States we have a very strong private sector, we also have arguably some of the world’s leading universities, like MIT where I am now, and Stanford and Carnegie Mellon who are doing outstanding work in these areas. Trying to harness some of those natural advantages is really the core of the US strategy. I am glad that other countries too are focusing on this area. Ultimately, as more people, countries and organisations invest in AI, it has the potential to benefit all of humanity. Some of them will undoubtedly come up with cures for diseases or ways of diagnosing cancer that we didn’t have before. We’ll have better self driving cars and predictive machines.
On US immigration policy.
Unfortunately, immigration policy is affecting the US in a negative way and that’s a point I made in the White House meeting. The United States is being perceived as more hostile to a lot of high skilled immigration than it was in the past. It’s making it more difficult for my own students to continue to stay here and study here and work after they finish their PhDs. It’s discouraging people from coming here in the first place. This is obviously a bad thing for the US and I’d say it’s bad for the world as well. Top researchers gain when they work together, shoulder to shoulder. If we make it harder for them to come together in places like MIT and Silicon Valley, I don’t think science would advance as rapidly as if we have free flow and people could join forces.
On an ongoing research project: Digital technologies and their impact on the earnings prospects of American workers.
What we’re finding is that digital technologies are making the pie bigger. They’re creating more wealth than the world has ever seen before. In particular, we are seeing more millionaires and billionaires than ever before in human history. On the other side, it is not as good. It is also creating inequality and leaving a lot of people behind. It’s that long tail I was describing with a lot of people struggling just to get by. Unfortunately, there’s no economic law that says everybody must benefit from these advances. It’s possible for some people to be left behind. It’s even possible, theoretically, for a majority of the people to be left behind while wealth is concentrated in a very small percentage of people. It doesn’t have to be that way. These are different outcomes that are possible. Technology is a tool and as we’ve stressed in our books The Second Machine Age and Machine Platform Crowd, these tools can be used to concentrate wealth or they can be used to create more prosperity. In fact, Andrew McAfee and I created something we call the Inclusive Innovation Challenge where we’re giving away $1million to people and organisations that are using technology to create more widely shared prosperity. We’ve identified hundreds of organisations that have entered the contest and in the Fall we’ll be awarding $1 million to the organisations that have used technology not just to create wealth and benefits but to create shared prosperity where many people are participating.
“If the crowd comes up with a better idea, how will you bring it into your core?”
Seems so obvious but most companies don’t understand it and that is that no matter how many smart people you have in your management and no matter how many smart people you hire, most of the smart people in the world don’t work for your company! Most of them are outside your company. Perhaps more important is that people outside your company think different. They have different ways from your culture. When they come at problems, they approach them in a way that is more likely to have a solution. If your people haven’t been able to solve a problem for a long time, maybe they’re looking at it the wrong way and somebody from a different organisation or a different part of the world will see it in a different way and find it is very easy to solve. There are platforms like topcoder and many others which you can tap into to solve those kinds of problems and it requires a change in culture. If managers want to be successful, they need to be willing to take ideas from the outside and the most successful managers have been doing exactly that and then integrate it into the organisation. There’s even a 100 or 1000 fold improvement in key processes when companies successfully tapped into the crowd.
On whether this is the single most important question for any business? — “If our competitors implemented a successful machine learning system for _______, we’d be in serious trouble.”
I think it probably is. Because the great general purpose technology of our era is the machine learning category of artificial intelligence and it is now able to do almost any task that previously only humans could do. In a paper I wrote with Tom Mitchell titled — ‘What can machine learning do?’ we describe a rubric that describes which kinds of tasks machine learning is good at and which ones aren’t. By applying that rubric to your organisation, you can see where machine learning can work well and, as we say in that questionnaire, maybe your competitors are applying machine learning and that might change the economics of your business very very fundamentally. We’ve seen it happen in a number of industries and whatever industry you’re in right now, there’s no reason to think you’ll be immune. Understanding the power of machine learning is a key challenge. A Harvard Business Review article which Andy (Andrew McAfee) and I wrote is the one I’d like to quote from now: machine learning is not going to replace managers but managers who know how to use machine learning will replace managers who don’t.
The views expressed above belong to the author(s). ORF research and analyses now available on Telegram! Click here to access our curated content — blogs, longforms and interviews.