Q&A with Erik Brynjolfsson on Disruptive Technology and the Digital Economy

Preview

Erik is a Professor and the Director of the Stanford Digital Economy Lab, who contributed his perspectives to our Charting Disruption 2022 Outlook. This interview with Erik has been edited for length and clarity.

There are so many disruptive technologies that are rapidly emerging today. What are some of the most exciting ones that you’re seeing?

Well, if you’re not excited, you haven’t been paying attention because right now we have a cluster of amazing technologies in biotech, digitization, mobile, but the one I’m most excited about is what’s happening in artificial intelligence.

AI is what I call a general-purpose technology. Like the steam engine or electricity before it, each of these technologies drives waves or cascades of complementary innovations that restructure the way business is run. Ultimately that drives productivity.

In fact, I would make the case that artificial intelligence is the most general of all general-purpose technologies because we use our intelligence to solve all the other problems in the world. If we can crack artificial intelligence, that would be the biggest change ever – bigger than electricity, bigger than the steam engine.

Why is it the case that we haven’t seen productivity really take off in the last 10 years since AI technology has been around?

One of the great paradoxes of our current economy was that we’ve had these amazing technologies and more and more of them coming online, especially in the past decade. And yet productivity hasn’t really grown much. In fact, productivity growth was about 2.8% in the 1990s and early 2000s. And for the past decade and a half, it’s been only about 1.3%. So less than half as fast as it used to be. Now, why is that?

Well, it goes back to the idea of general-purpose technologies. A GPT, a general-purpose technology, is powerful but it also requires a lot of transformation of business processes, new products and services, new skills, and those all take time.

With the steam engine, it was about 40 years before we saw big productivity growth. With electricity it was about 20 to 30 years. I’m hoping it will be even shorter with AI and other related general-purpose technologies. But until we had that business transformation to really harness the technology, we’re not going to see big productivity growth.

I am confident that productivity is going to surge. In fact, there’s some evidence it is already beginning to surge. In the next decade, I think productivity will be far better than the congressional budget office predicts.

It seems like any smart business out there should just snap their fingers and put in place AI or robotics anywhere they can. But what are some of the hurdles to actually implementing this technology?

You know, I wish it was as easy as snapping your fingers to just put these technologies in place. But that’s very rarely the case.

It’s not enough to simply buy a shiny new piece of technology or a new software system. You also have to re-skill your workers. You have to change the way you interact with your customers. Rethink your supply chain. All these changes take time.

When Amazon reinvented the bookstore, they didn’t simply replace a human cashier with a robotic cashier that would have been kind of boring and not all that productive. They reinvented the whole system and that kind of creativity is hard to come by, but we’re seeing it play out in a lot of businesses over time.

Are there specific areas that you found are squarely in the sights of machine learning versus some other areas that AI is very far off in terms of really being able to be applied?

Absolutely. Some broad categories that we found were, for the things that were routine and repetitive, machines could do them better. On the other hand, things that involve a lot of creativity like entrepreneurship, scientific discovery, art, were very hard for machines to do in a serious way.

Also, interestingly, there are a lot of tasks that involve high emotional intelligence – childcare, nursing, coaching, teaching. Those often also were difficult for machines to do because they involve a human touch and a human interaction.

When we looked at the economy more broadly, what we found was that almost every occupation had some of their tasks that were suitable for machine learning. However, in no occupation did machine learning just run the table and do all of the different tasks they had to do.

It’s not as simple as just automating tasks. Instead, what we find is that you have to rethink how tasks are put together. Simply taking a job that a person is doing right now and trying to replace it with machine learning rarely works. Instead, you need to combine humans and machines into kind of a hybrid or centaur system, and that’s where you get the most value. The companies that figure that out are going to pull ahead. We’re already seeing the top 10% of companies pull ahead from the rest of their competitors and I expect that to continue over the next decade.

Are there challenges in measuring the economic impact of the digital economy?

Yes. One of the great ironies of the digital age is that we have more information about the economy than ever before, but our economic measures are actually missing more and more of the value creation. Digital technologies often have zero price. Wikipedia, you don’t have to pay for that. And a lot of other goods and services. But GDP only measures the things that are bought and sold in the economy. So if something has zero price in the GDP statistics, it’s almost invisible. It’s like it doesn’t count. Now we’re still getting value from these goods and services. They’re not just showing up in our GDP statistics and our productivity statistics. That’s why we’re going to have to reinvent the way we measure the economy, just like the way we reinvent the way we produce goods and services.

How do you overcome those challenges if it’s not showing up in today’s largely accepted economic metrics? How do you propose that that changes?

Well, if you go back to the invention of the GDP framework, GDP was really not a measure of wellbeing. It’s just a measure of what the cost of production is and the value that it’s getting for companies.

If you want to measure the value that consumers are getting, you have to look at a concept that economists called consumer surplus, which is how much value you’re getting compared to what you have to pay. And as the price goes down, GDP goes down, but consumer surplus goes up. And for most of us, we care more about the consumer surplus. So when you replace encyclopedia with Wikipedia, that’s less GDP, but more value.

We are right now in the process of measuring the consumer surplus for thousands of different goods in the economy, and what we’re finding is that there’s a staggering amount of consumer surplus from these digital goods – hundreds of billions of dollars of value being created. It’s missed purely from the GDP statistics. We see a whole parallel set of accounts alongside traditional GDP.

We want to introduce GDP-B, where the “B” stands for benefits. Then, we can see not just how much money we’re spending, but how much value we’re getting. As the economy becomes more and more digitized, it will become increasingly important to look at this GDP-B.

So Eric, what is digital resilience and how do you measure it?

The pandemic was a huge shock to American businesses and some companies managed it much better than others. We call that digital resilience. There were some companies that were able to plow ahead and actually increase productivity, profits, customer satisfaction and gain market share from their rivals.

We dove in deeper to figure out what were the differences between the digitally resilient companies and the others. It had a lot to do with their digital infrastructure. They had more cloud, more data science, more technology, and more sophisticated skills in their workforce.

What we’re seeing now is that some of the companies that were left behind are rushing to try to catch up in terms of digital resilience and they’re investing faster than the leaders in an effort to close that gap.

Did you see these differences and digital resilience ultimately affect the bottom lines of these companies?

Digital resilience had a huge impact on the bottom line of these companies. You could measure it in profitability, but also in other indicators like productivity, customer satisfaction and overall market value. The leaders were already ahead, but during the pandemic they increased their lead and the top 10% of firms have pulled away from most of their competitors.

What does digital resilience look like after the pandemic?

I think we’re seeing a lot of the companies that struggled, trying to catch up with their leaders and they’re realizing that it doesn’t pay to skimp on their digital investments. So those companies are actually investing more than they did before. And what we’re seeing is the potential for all companies to become much more digital than they were before. One of the interesting questions is as other shocks hit, not just a pandemic, but earthquakes, hurricanes, and other types of shocks, whether the same kind of digital resilience that worked in 2020 is also going to help them with these other shocks. I suspect it will.