3 Takeaways - The Exponential Age: How Accelerating Technology is Transforming Business, Politics and Society with Tech Seer Azeem Azhar (#109)

Episode Date: September 6, 2022

Technology is advancing at exponential speed, and humanity is having serious trouble keeping up. Azeem Azhar, a tech seer who has founded and sold four companies, shares his unique insights into what he calls the “exponential gap” and its impact on business and society.

Welcome to the Three Takeaways podcast, which features short, memorable conversations with the world's best thinkers, business leaders, writers, politicians, scientists, and other newsmakers. Each episode ends with the three key takeaways that person has learned over their lives and their careers. And now your host and board member of schools at Harvard, Princeton, and Columbia, Lynn Thoman. Hi, everyone. It's Lynn Thoman. Welcome to another episode. Today, I'm excited to be with Azeem Azhar, who has founded and sold four tech companies and is the author of The Exponential Age. I'm excited to find out why Azeem believes the rate of technical change is accelerating and how this will transform business, politics, and society.

Welcome, Azeem, and thanks so much for our conversation today. Lynn, thank you for the invitation. Azeem, I loved your book. I really enjoyed reading it. So thank you. Thanks so much. Azeem, why do you think we're in a period of accelerating technical change? Well, because the data shows the rate with which technical products like AI or electric vehicles or the number of genomes that are sequenced is increasing. And it's increasing because the underlying technologies that deliver those are getting rapidly more cheap because we're learning how to build them better. And so in a

sense, it's not so much a belief as it is an observation of the evidence. I'll give a simple example of that, which is that the amount of time that it took the network, social network TikTok, to go from being launched to a few billion users was so quick that when I started writing my book, TikTok was just a tiny little play thing that no one cared about. By the time I finished writing my book, it was a major national security threat to the US. Now, even a network like Facebook, which grew really, really quickly, simply did not grow as fast as TikTok in that it was only 10 or 15 years ago. So it's really grounded in the evidence and what we see. And you believe that it's individual industries, such as computing, energy, biology, and manufacturing that are all experiencing

really rapid change, but it's essentially the layering of all these different industries or technologies upon each other or interacting together that is further accelerating change? Yeah, I mean, it's a complex system with lots of feedback loops. And I think of it really more as what economists call general purpose technologies. So these are technologies that can be applied more broadly than really narrow technologies. So something like a sewing machine can really be used for one thing, which is sort of for sewing, whereas something like electricity can be used in an operating theater, on a plane, in a video game, to light our homes. And what's interesting about the period

we're in now is that there's a set of technologies that are sort of generally applicable in different domains, computing being a key one, but also these new technologies of energy storage, like sort of lithium batteries, but also the new technologies of manipulating biology and the stuff of life. So that is reading the biological code through sequencing, but also reprogramming that through cellular programming or through writing synthetic bases. And that even within the field of manufacturing, additive manufacturing or 3D printing, which can be used in many, many different types of uses and industries, all of these core technologies are general purpose that are getting dramatically cheaper

every single year. And that is on a compounding basis. And by dramatically, I sort of chose the arbitrary number of 10% cheaper every year compounding. And that very quickly means that something that was expensive or is expensive today will in a few years be very, very cheap. And if I can give just one example of that, the first transistors were developed in the mid 1950s. And the transistor is sort of the most important single element in a digital computer. It's a kind of the thing from which bits flow. And when IBM bought a bunch of these in 1958 from Fairchild Semiconductor, they paid in equivalent $20, $21, about $1,200 to $1,400 for each one of them. And they bought about a hundred. In today's iPhones, you will find tens or hundreds of billions of transistors. And in fact, the cost of a transistor has fallen from 62 years ago, from about $1,500 to

tenths of a billionth of a dollar. So you could get 100 of them for a millionth of a dollar, even in this era of inflation. And so that is what that kind of compounding cost decline ends up doing. And what I observed in my research was that this was happening in computing and related fields, but it was also happening in solar power and wind turbines and batteries, and it's happening in genome sequencing and so on and so forth. And that's the really distinct nature of this period of time. And one of the trends that you talk about is how additive manufacturing means that businesses no longer need to rely on low labor countries to produce goods and ship them all the way to the wealthier developed countries, but that they can manufacture

them much more inexpensive locally. Yeah. Additive manufacturing, which is the, I don't know if I can say this, the ugly stepchild of exponential technologies, because it's been 3D printing and it's the one that has taken the longest to really take off, but it's starting to. It's so powerful because essentially you separate out the feedstock that you require to print the thing you want from the plans, which are just IP. And so you could start to produce in quantity the components that you need in finished manufacturers locally rather than shipping them. And then you could start to do final assembly, which might be by humans or it might be by robots or humans and robots working together locally. It's not going to work for every class of product by any means. You can imagine really sophisticated products like

semiconductors for which it's going to be those really complicated manufacturing processes will require much more thought than that. But you can start to see it working for many other classes of product, even at much larger scales. So one industry that's surprising is doing quite a lot of 3D printing is construction, where homes like kit homes are being 3D printed and then just sort of assembled very quickly. So the real question is, can the exponential trend that brings the price of this technology down while taking its performance up, kind of continue for long enough for us to use 3D printing in more mainstream applications. Azeem, how do you see the trend to win or take all

and greater concentration? It's a challenging set of dynamics within exponential age firms because they, in a sense, they rely at the heart on data and that data benefits from more data. So there is a moment where you have increasing returns to scale that can accelerate. And that sort of expresses why booking.com is the dominant travel booking service by a long way in the world, or Google is dominant in its industries. And what you see in markets that have been touched by the hand of exponential technologies is that we can name number one and we can name number two, we don't even know who number four

is. And so it breaks our traditional sort of Marshallian view of like what markets end up looking like, which is you get this kind of managerial complexity that it's really hard to manage these big companies. They get sort of bogged down into treacle. And so companies don't get much bigger than a certain scale. What we've seen instead is a set of patterns that deliver increasing returns to scale. And that's why you end up with companies whose inherent heartbeat is that they're going to be bigger. And then they can use traditional techniques like pricing or locking up supply to cement the advantage that these network effects have given them. What I think about this is that a lot of our thinking

about what makes industries work, which is that they have to be competitive, that market entry has to be easy, is market entry has to be easy, is sort of predicated on easy competition and diminishing marginal returns. And if those things disappear, then our assumptions about the rules that we need to manage those industries to make them competitive and sort of pro-competition, pro-consumer, pro-economic welfare probably need to change. And it's really interesting in the year or so since the book came out, there has been a change, a shift in the FTC with a new boss, Lida Khan, who I talk about in the book, coming in and trying to rethink what dominance means in these markets, because the sort

of old test of what monopoly was doesn't seem to work. How powerful are these platform companies? How powerful is Facebook, Mark Zuckerberg, and why is that a problem? accountability, as well. So in any, I think, in any polity where you've got different organizations and agents, you want to have those checks and balances in place, whether they're sort of formalized as they were until recently in the US, if I may make a slight mordant joke about the state of your constitutional law at the moment, or it's done informally. And so I look at this as an issue of these companies getting more and more powerful, and then starting to have almost state-like functions, like quasi-state-like functions or impact in certain domains without having the checks and balances and the democratic accountability to pull them back. is a sort of negative way, which is, are they really looking after our interests? And can they ever, given the fact that they're companies with sort of shareholders and board of directors and so on. But on the other hand, there are things that they can do in this new environment that

states can't do. And I'm not sure one would trust states to be able to do that, given where the sort of political layer lies. And a really good example on that latter point is that when there are cyber attacks and these sort of viruses and worms start running around, they spill across national boundaries. And no national government has the technical capability to intervene at that point. And you're reliant then on these sort of platform companies like a Google or a Microsoft to identify the threat and to attenuate, mitigate, and shut down the threat and issue a fix. So I think we should be concerned with any situation where these companies become

as powerful as many of them have become. And it's not because they're putting other industries out of business. It's not because we don't like them as characters. It's not because like Apple with respect to privacy or Microsoft with respect to sustainability, they can't be really, really good actors in it. It's that ultimately they're incredibly powerful and there are no checks and balances and they're more powerful than we perhaps expected. And that for me is the discussion that we need to have around the scale and capabilities of these firms. Let's talk about jobs and employment. Can you tell us about the trend of having more valuable companies with fewer employees? It is such an apparent trend. I mean, if you think about the scale of General Motors at its

peak employment was around a million people. And you look at companies like Apple, which are producing more revenue, more profit, worth much more with a sort of a 10th roughly or a fifth of the peak GM headcount. And the reason for that is ultimately that around automation and productivity that drives a real increase in productivity in those workers. And you can see it across many, many data points. When you look at the amount of value that gets created per individual employee in subsequent technology firms at similar stages, it continuously increases. So Instagram, which Facebook acquired, had fewer employees at a similar level of development to Facebook, WhatsApp, fewer still. And the reason is that we're able to shift a lot of the daily functions into automated bits of software, whether we access them through

the cloud and what's known as SaaS products. So instead of having to build your own billing system today, you just go to Stripe and it's one line of code to drop in to have a full billing system. And that then your employee can move on rapidly to sort of the next piece of work that they have to do. And I have a great story with my last startup, where we had a guy who was hired to tend the servers. And we had about, yeah, I don't know, half a dozen servers at the time, the details are in the book. And as we were growing, we had more and more servers coming on stream. And what he was able to do was reduce his five day a week looking after servers and

patching them and making sure they had the latest security software and so on to half a day a week. Because through automation, he made that part of his job more productive. Now, because we were a fast-growing company, he then got a mentor and was able to move up into software development, which was sort of higher salaried work while maintaining his own responsibilities. And I think that that kind of productivity shift, which requires an investment both in technological capital and in human capital is something that we have seen in some of these exponential firms. The better or more advanced the tech gets, do you think we have more or fewer jobs? Amazon is one of the most automated companies in the world

with one robot for every four workers. As you point out, has been on a hiring spree. What's going on? In the long term, we'll have more jobs. We may redefine what jobs are, but we'll have more jobs. Successful companies grow and growing companies need to hire. The companies who can grapple with breakthrough technologies and have the leadership and the management to implement those technologies will be well-run companies with good quality execs. So they'll be able to out-compete companies that don't make those choices. So if it's 1921 and you decide not to open a car dealership, but rather to open a blacksmith's, you've been a poor manager and your blacksmith's business is not going to grow as quickly as Penske Automotive has grown too, right? So I think a lot of it is, I would argue, down to sort of traditional

good managers, good companies perform better. The fact that they execute through a strategy of automation is, by the by, they could have been executing through some other strategy. And what you tend to see is that those companies grow. What longitudinal data shows is that where there are declines in employment, it tends to be in the competitors that did not compete as well because they didn't use the technologies as well. Overall, you believe there's a gap in our institution's ability to change and the accelerating speed and impact of new technologies. Can you summarize the implications? Yeah. I mean, I think that if governments don't step in, in some sense, and it's not just governments, it's other institutional structures, what happens is that you essentially allow the forces to run amok, right? You allow companies to get really, really big. They can dominate and dictate

the shape of industries. You construct a very, very febrile geopolitics where there is lots and lots of incentive for these low cost gray behaviors like disinformation and cyber attacks. You allow companies to look at the employment question and purely use the technologies of monitoring and algorithmic management to drive down what it means to be a human in the work experience. And humans instead start to look like boxes in a spreadsheet, which is a kind of divisive way of taking things. In that, I think what you end up doing is you end up moving away from any kind of notion of social welfare. And that's what happens if the exponential gap doesn't get closed.

And that gap is really about the fact that the ways in which governments need to think about their role, the ways in which they think about competition policy, the ways they think about what is it for markets to work effectively, or how should workers be able to relate to their employers when the power is suddenly shifted to the employer. I think all of those things become problematic. And in order to close that gap, you then have to be a bit more catalytic and a bit more directional. I don't think it means that you get less speed of technological innovation. I think what you get is a sort of better,

more directed, more pro-social styles of that, which still creates as many longstanding economic opportunities as any other period. Before I ask for the three takeaways, is there anything else you'd like to mention that you haven't already discussed? What should I have asked you, Azeem, that I did not? One question is, is this like previous periods in history, right? Are there periods that we could walk back to? And I think that there is a period of time between specifically the 1890s and 1920s, where we saw lots of institutional change and technological change and geopolitical change that has some similarities, but rather it is, there are patterns that we can look at and you

have to, I think, dig a little bit deeper than the obvious parallels to try to see what those patterns were. And I think asking that historical question is really, really relevant because it starts to put some shape on also questions of timing, questions of politics and trade and diplomacy, which are becoming increasingly important. What are the three takeaways you'd like to leave the audience with today? Takeaway one, I think it really is a time to rewrite assumptions and revisit priors about what the world could look like, what the technologies are and how they need to

be governed and how quickly we should be rolling them out, which in general, the answer to the latter is as fast as we can, because they're going to be a necessary part to tackling climate change. The second takeaway is that despite that, technologies will scale and move faster than we expect, even big heavy ones. So things like the rollout of solar power around the world has accelerated, exceeded everyone's expectations. In Norway, it took seven years for 3% of the cars sold to, there are people who will be less powerful. Countries will be less powerful. Companies will be less powerful. And that will create instability. That lens of instability will play out and will contribute to that. And I really think that there will be a new framing of what the global order is in the same way that we used to say, well, before Bretton Woods or

before the oil crisis and after the oil crisis, before World War II and after World War II, before the global war on terror and after the global war on terror. I think there'll be a kind of before exponential age, after exponential age moment where the way in which the order of global affairs gets organized, discussed and negotiated will fundamentally change. And that would be my third takeaway. And we certainly need that. Well, we're going to get it. So need it or not, it's coming. Azeem, this has been great. I really enjoyed The Exponential Age. Thank you. My pleasure, Lynn, and Facebook. Note that 3takeaways.com is with the number three. Three is not spelled out. See you soon at 3takeaways.com.