Q&A: 'Rise of the Robots' Author on AI's Talent Crunch and the Future of Work

It may take some major technological breakthroughs before robots are able to think creatively or handle complex human interactions. But futurist Martin Ford, author of the New York Times bestseller Rise of the Robots, says that won't stop current automation and machine learning technology from changing the face of the labor market in the next couple decades—and office workers shouldn't assume their jobs are safe. The nature of the jobs that will be replaced has less to do with training and salary and more to do with how routine and predictable the work is, according to Ford. That said, AI is nowhere near the point where it could do the work of creatives or occupations that require wide varieties of tasks. Meanwhile, people who understand the ins and outs of AI systems have never been in higher demand—even if they lack a formal education in the subject, Ford says. Ford, who's publishing his next book on AI later this year, spoke with Adweek about the opportunities and limitations of AI, the shortage of tech talent and which jobs are most vulnerable. The following has been edited for length and clarity. Adweek: AI has become such a buzzword in the business world that it seems to have stretched the real definition of the term. How do you separate hype from substance? Martin Ford: I have a practical view of it. If you look realistically at what's going on, it's pretty clear that a lot of routine, predictable tasks are going to be subject to automation. The hype gets carried away about superintelligence and really advanced applications, and the idea that we're going to have fully autonomous cars within five years. I think it's easy to see areas where things are going too far, but at the same time, it would be a mistake to underestimate the transformation. I tend to think of it over a 10- to 20-year time frame. I'm not thinking in terms of what's going to happen next week. But I do think there's enormous potential for a huge transformation. What needs to happen in terms of data and computing power to realize the changes you talk about? We can assume that a lot of things will be automated, and the nature of work will change just by assuming kind of an extrapolation of what we have already. We don't have to really assume great breakthroughs in artificial general intelligence. If that does happen, that would create an even bigger disruption than what I'm imagining. Really, what I'm looking at is just sort of a progression based on the things we see already. I mean, we're already seeing algorithms that can outperform radiologists. This is happening already so within 10 to 20 years there's going to be a much bigger impact. What should be done to solve AI's talent shortage? I think there is a lot of interest in solving that talent problem and a lot of investment there. It's one of the few professions where you can really make a lot of money, where the entry barrier really isn't that high in terms of formal credentials. If you want to be a doctor or lawyer, you've got to jump through lots of hoops, but you can teach yourself machine learning. The shortage is enough now where a lot of companies will look at you even if you don't have formal credentials. What are some of the biggest misconceptions about automation and AI replacing jobs? There's not necessarily a correlation between what can be automated and how much skill or education it takes to do that job. The radiologist, for example, a computer can do that, but a computer can't drive a car. Compare the training between a human driver and a human radiologist, and it's actually the reverse. There is still a bias to assume that it's going to be blue-collar jobs and people who don't go to college, and to some extent, that's true because there is some correlation between not being skilled and doing more routine, predictable things, things that can be automated. But there are many examples where it's not true. The person who cleans your hotel room, for example. Think of the amount of dexterity and problem-solving in a completely unpredictable environment that requires. When it comes to AI in ad tech, is there a way to bring more personalization without giving people the sense that their privacy is being violated? Are we going to see more pushback as these systems get smarter? I think people have pretty much accepted personalized advertising in the online world. I think if we see a backlash, it's likely to be when these technologies get deployed more in real-time environments—for example, if facial recognition or location tracking is used to display personalized ads in real-world environments (inside an Uber for example).