He Has Covered Tech for 39 Years, and What Still Surprises Him Most Is That Starting a Company Became a Way of Life

You may not know much about this veteran New York Times tech reporter, but he has been "tracking" artificial intelligence, driverless cars and virtual reality for a very long time. Talking with him is fun, because in the end he would still rather be the one asking the questions.

2016 marks the 39th year that New York Times reporter John Markoff has covered technology.

When Markoff, who grew up in Silicon Valley, started reporting in 1977, Apple was still a niche toy and computers were IBM's business.

In 1988 Markoff joined the New York Times. By then Microsoft and Intel had put the personal computer in every office and home. In the nearly 30 years since, Google, Apple and Facebook have risen one after another, and the phone has replaced the PC as our main personal computer.

With the smartphone market saturated, the tech industry is already looking for its next wave. Artificial intelligence, robotics and virtual reality, which people thought were just around the corner four or five decades ago, are once again the fields tech companies are racing to win.

Markoff won a Pulitzer Prize in 2013 for his reporting on automation and the future of human work, the subject of his book Machines of Loving Grace.

Earlier this month, QDaily met Markoff at a café at Tsinghua University to talk about where technology stands today. How close are we, really, to the AI, virtual reality and robots that people foresaw four or five decades ago? And where will the tech industry, which has already changed the course of our lives, go next?

The next trend? Virtual reality is still five or ten years from going mainstream

QDaily: Everyone is debating what the next wave will be. Which field do you like best?

Markoff: I'm not a prophet. All I can say is that I'm looking forward to AR, because the phone is certainly not the final form of the user interface. Something will come after it. But as with driverless cars, I don't think it arrives next year. It may take another five or ten years.

I was the second reporter in Silicon Valley to see a Magic Leap product. What I saw wasn't a finished product, just a lab test, but the experience alone stunned me.

For example, they can project a cup like the one on your table into my field of view, and unless I reached out and poked it, I might not realize it was fake. That's how sharp it is. They blocked out the background shadows and made the image ultra high-definition. I had never seen a demo like it. It was better than I expected.

But I also know they have plenty of problems. They don't know how to bring this technology to consumers either: what form it should take, who the users are, how to market it, what it's for. And it will probably be expensive. Still, an interface you can interact with, simulated in the real world: I think that's really cool.

QDaily: Microsoft is also working on AR. Between HoloLens and Magic Leap, which has the edge?

Markoff: Microsoft is the more interesting company. It has lost its monopoly. It still collects Office license fees, but it doesn't yet have a new category of product that sells in huge numbers, and that's why it's building HoloLens. I tried HoloLens, and before I compared it with Magic Leap I thought it was okay. Now I think the resolution is too low, about VGA level, and the gestures are too complicated and feel unnatural to learn.

The best thing about Magic Leap is that I can spread my hands apart in mid-air and pull out a control panel anywhere in the space around me. I could sit down anywhere and start working: wearing Magic Leap, I can mark out a patch of air as a display that only I can see. It's very cool.

QDaily: Back to AR. I'm curious why you didn't say VR. Do you think VR has problems right now?

Markoff: The VR experience still falls well short. First and foremost, it's uncomfortable to wear. I get dizzy easily. I tried VR headsets very early, and afterward I stayed dizzy for more than an hour. The founder of Magic Leap has said the dizziness this kind of VR causes could do permanent damage. I've also used Oculus, and it made me very uncomfortable, even though I only ever wore it for a short while.

But I understand why investors and others are excited about VR. The product is very appealing. In games, for example, VR would greatly change the experience and make it far more immersive.

Every year someone asks whether this is the year of VR. What I can say is that it will take at least ten years, maybe longer, for VR prices to fall to an acceptable level. At $500, VR headsets are not going to become the next mobile wave any time soon.

(So how low does the price have to go?)

I'd say $100 or less, the price of a Christmas present for a small child. Only then might VR headsets spread. I think investors, and everyone counting on VR to take off quickly, are a bit too optimistic right now.

QDaily: Which uses of VR do you find most appealing?

Markoff: Changing how we watch films and live broadcasts, for example. A friend of mine in Israel is building a service that would let ordinary people use VR to watch matches at the 2022 World Cup in Doha from different angles, with multiple viewpoints around the stadium. I find that kind of use very appealing. Games are pretty good too, but VR games are, after all, for gamers, because only fairly high-end computers can run VR headsets. That means carving a small slice out of the game console market.

QDaily: That makes me think of Intel. It is making smaller chips with more computing power for VR, the Internet of Things and drones. For decades Intel has run its business on Moore's Law, under which chip performance doubles and price halves every 18 months, and so has Qualcomm. Now that the smartphone market is saturated, these smaller, more energy-efficient chips seem able to do a lot, yet IoT, wearables and VR have not produced the next product as universal as the smartphone. How do you see this?

Markoff: All I can say is that Moore's Law has slowed down. Computers used to be the size of a room, but under Moore's Law chips shrank by half every 18 months while getting better, so computing devices were bound to keep getting smaller. That gave us the personal computer, and later the smartphone. That was all Moore's Law at work. But in the last year or two, at the academic conferences I've attended, the people who study Moore's Law and the people who follow the chip industry have all been saying the same thing: the law has slowed down.

Cheap labor and robots are not in conflict

QDaily: After the smartphone, what do you think will change human life?

Markoff: You know, I'm just a reporter. Coming from Silicon Valley doesn't make you a prophet. When you ask reporters to make predictions, they always get them wrong. I can only talk about what is already happening.

Everything around us now is a "machine." That's not a pejorative; some people call them "robots." The more people interact with them, the more human they become, or the more they become stand-ins for humans, which is morally very controversial. Like Siri and Cortana, they don't necessarily have bodies; they may be just algorithms. Yet people like them, simply because they are useful.

What we can foresee is that these robots will change human life a great deal over the coming decades. I don't know how long it will take, but a friend of mine in Silicon Valley puts it this way: never mistake a clear view for a short distance.

I think that's a good way to look at it. In the short term the change is slow; in the long term it will be dramatic. But you may not feel it within five years.

If you told me that in 2025 a robot or a driverless car could drive me from San Francisco to Palo Alto for dinner, I wouldn't believe you. First, on the technical side, driverless cars are still far from recognizing images and driving safely in rain and snow well enough to keep us safe. Second, on the legislative side, the safety and liability questions raised by putting driverless cars on the road slow things down even more.

QDaily: If everything around us is a computer, are our lives already controlled by machines, or rather by algorithms?

Markoff: That depends on how you define "controlled." We are certainly surrounded by algorithms. The algorithms in phone apps keep steering us: which restaurant to go to, whom to date, even whom to marry. You've probably heard of Facebook's experiment, which was very controversial. They showed people different information and then watched how it affected the way they voted. That is using algorithms to influence the democratic process, and it worries me.

For tech companies, the algorithms are constantly tuned with our behavioral data, yet in the end they become the companies' commercial products. Tech companies such as Google always want people to be able to use them everywhere. So they watch me and collect my data at the same time, and the data and the algorithms built on it become their property. There is a bill in the United States that aims to regulate these companies' algorithms, but I still think that's very hard, because how do you define who owns this behavior and this data?

If I search for something on Google, who owns my behavior? Me, or the company? As things stand, it all belongs to Google. Baidu is the same: it claims ownership of your behavioral data and content. But I think the debate over this question, and the fight it eventually leads to, may be a turning point for technology.

QDaily: Right. Another hugely controversial topic is robots taking human jobs.

Markoff: When I started writing Machines of Loving Grace, I was actually rather pessimistic. In 2010-2011, when people complained that robots would leave humans with no work, I thought they might be right. But once I started researching the subject, I found the alarm was misplaced.

Technology limits particular occupations, but it doesn't limit work.

Technology also creates new jobs. AI may change where and how you work, but it won't make work disappear. Take SEO. Before Google became a big company, who would have imagined that line of work? Tens of thousands of people in the United States do it now.

2011-2012 was when Americans worried most about losing their jobs to machines. Now the economy is somewhat better, so the worry has faded. Interestingly, McKinsey and MIT have done similar studies, and neither believes that ten years from now the impact will be so large that there is no work left for people.

QDaily: But it sounds as if technological progress will make it easier for robots to replace certain kinds of work, such as assembly-line workers and blacksmiths, the manual jobs.

Markoff: Actually, my view now is that the aging of the human population drives robotics more than technological progress itself does. That's especially true in countries aging quickly, such as Japan, South Korea and China. The United States is a little better off because it is a nation of immigrants. Europe's problem is as serious as China's.

For the first time in human history, there will be more people over 65 than children under 5. By 2150, the number of people over 80 will be twice what it is today, and before the end of this century it may be seven times today's. That is what matters most.

China's labor force will shrink sharply over the next decade. The two-child policy is no substitute; people simply want fewer children. So you need robots, and that is a very large market. A transition has to happen too.

QDaily: Most of China's population works with its hands, so being replaced by machines sounds rather frightening.

Markoff: Yes, but don't worry so much about technology putting people out of work.

Technological development has actually had a smaller effect on human society than we imagine. Take productivity: it looks like this shape, not that shape. Technology hasn't had a dramatic effect on human work, and economists are still fiercely debating how much it has raised productivity. That makes me wonder: how big an impact have computers really had? It isn't clear.

For example, at the same time as people worry about jobs being replaced, the US labor force participation rate is at its lowest since the 1970s. But that's because my generation has started to retire, and the shrinking working population actually has a bigger effect than technology does. So it's really very complicated.

So in the end it's not as frightening as you think. A commonly cited example is the bankruptcy of Kodak, with its huge workforce, and the rise of Instagram. Did Instagram really replace Kodak with an algorithm, 30 employees and one product?

No. Kodak led in chemical film for many years; Instagram is a product of the mature internet. It isn't a replacement. It is one of the more than 2.4 million new jobs the internet has created. Fuji was a Kodak competitor too, and it's still alive. Kodak itself made some serious strategic mistakes, and those also led to its failure.

So it's not as simple as robots replacing jobs.

Why did artificial intelligence take so long to arrive?

QDaily: Let's talk about artificial intelligence. Tech companies are all competing in AI now. Your book lays out an interesting fact: the AI and robots we have today are what Silicon Valley was already working hard to build in the 1950s and 1960s. Why did we wait so long, only to get AI to where it is now?

Markoff: Back then there were two groups with different values, attacking the problem from different angles. One, led by John McCarthy, was the "artificial intelligence" (AI) camp. They focused on building up the computing power of AI, and many of them simply loved robots and believed that building robots smarter than humans was where history was heading. That was also the strongest current in Silicon Valley's computer industry at the time.

The other was the Stanford Research Institute, led by Douglas Engelbart. What they did was called "intelligence augmentation" (IA): they wanted computers to help humans, as an extension of human intelligence. Engelbart's vision contained early forms of the internet, word processors and other ways of interacting with computers, though more people know him as the inventor of the mouse.

Researchers in both camps saw the personal computer as a niche thing, like the hippies, the bohemians and the rest of the counterculture. As it turned out, driven by Apple and Microsoft, the personal computer became the most important technology product of the following decades. And "artificial intelligence," which people once thought would soon work well, did not work well for decades. Only in the last few years has AI come back into public view.

But the products we use today bear the marks of both camps. Take Andy Rubin (Editor's note: founder of the Android operating system and the first architect of Google's robotics division). He believes robots replacing human work is the only direction. I think he's quite optimistic. The change may come more slowly than he imagines, especially when you set out to build a robot that replaces human work. They are improving the technology, but I don't think they will solve the problem in the short term.

Apple's Siri, on the other hand, uses AI to help people, which is using AI to augment human knowledge and intelligence. So I think the two camps may join forces in the future. But in the end, the two camps still represent the values and responsibilities of today's product designers.

Put plainly, robots are the designers' responsibility: did the designers, while building the product, think about the consequences the technology might bring? I know some didn't. As I said, AI only became mainstream again in the last five years, so it's still too early to say what will happen. Over the next decade we may see how these technologies get used, and the outcome of AI vs. IA will emerge. But I'm not sure it will be a very good outcome.

QDaily: Why do you say that?

Markoff: In my country, economic value dominates. I'm not entirely optimistic because I'm not entirely optimistic about how both camps' AI will be applied over the next decade. There are good parts, of course, but we are also seeing some very dangerous tendencies. You can now use AI or IA to design a new weapons system that lets the decision to kill bypass human judgment. More broadly, more than a hundred countries are now building smarter nuclear weapons systems, and that may not end well. If there is a war in the future, guided by AI, it will certainly be even more brutal.

QDaily: So do you think the superintelligence that Bill Gates and others worry about, one that acquires human intelligence all at once, will become reality? (Editor's note: superintelligence needs no learning process; once such an algorithm appears, it can instantly leave humans behind.)

Markoff: I'm not too worried. As for this so-called AI that can learn the entire internet at once without having to grow up, research has already shown that such AI still needs time to learn.

QDaily: Do you see any company with an edge in AI, or in intelligence augmentation? Is it Google?

Markoff: In a way, Google is a lot like Microsoft. Microsoft's power came from holding the chokepoint of information, that is, the means of controlling it. On MS-DOS, to find a piece of information you had to look on a particular drive. Google is very similar. Google controls the gateway to information, which is its search box. The only difference between them is that with Google you can type in words directly.

Microsoft's monopoly came from its control of information in the PC era. When the internet arrived, Google took the next gateway to information, and Microsoft's power passed to Google.

Google still makes most of its money from desktop search ads. But the trend is clear: tech companies like Facebook can pose a big threat to Google, and that has been happening for five years. And this is while nobody has found the next gateway to information.

QDaily: Five years ago, what did you think the next gateway would be?

Markoff: At the time I thought it would be mobile search, but now it doesn't look that way. So Google is placing all kinds of bets. What I'd say is that placing a bet doesn't mean you'll win. I'm very optimistic that some company will find the next gateway to information.

QDaily: What about Microsoft? What do you make of Xiaoice, the chatbot Microsoft is building?

Markoff: Microsoft's AI experiment in China still surprises me. Xiaoice has done quite well in China. It reminds me of the movie Her, in a simpler version. She is very entertaining and keeps you company, and people spend time with her, even though she doesn't have much practical use yet. In the US she would seem rather intrusive, but a Chinese friend of mine at IBM tells me Xiaoice suits Chinese culture quite well.

It is probably Microsoft's most interesting experiment in recent years. In China they pick up questions on all kinds of topics and reply in real time, in language that is very local and very entertaining, and people are happy to play with her. An interesting contrast: I know the developers of Siri, and they designed Siri to have no entertainment features at all. They did that on purpose.

But I believe Apple will come back and build this companionship into Siri, because that's how it becomes a better assistant.

This won't stay an experiment or a toy. My guess is that Microsoft is testing future ways of interacting, and interfaces, that suit people better, and Xiaoice is just one entry point for collecting data and improving what lies behind it. She will eventually become a useful tool, just not yet.

QDaily: After all your years in the business, which change has struck you most?

Markoff: What struck me most is how starting a company became a way of life! Young people in Steve Jobs's day started companies too, but they were a minority. The atmosphere of Silicon Valley's startup and venture scene has simply changed too fast.

Growing up there, I saw Silicon Valley as a place that thrived on artificial intelligence, computers and semiconductors. But the last wave made it less like Silicon Valley, with a lot of star-making. I mean the social network movement of ten years ago, which, to be honest, I didn't much like.

What makes me happy is that in recent years people have been returning to work driven by technology: autonomous driving, artificial intelligence, virtual reality, augmented reality... Honestly, I'd still rather sit where you're sitting now and be the one asking the questions.