We’ve heard so much in the past year about technological change, the “Fourth Industrial Revolution,” and how everyone from Elon Musk to Steven Hawking is afraid of robots taking over the universe. What’s going on?

Society is trying to come to terms with the increasing pace of technological change and this exponential rise of new technologies, and put those concepts into understandable frameworks and metaphors. Calling what’s happening right now the Fourth Industrial Revolution brings to mind previous patterns of change and side effects—especially relating to jobs and employment.

Current discussions about robotics and artificial intelligence (AI) generally involve extrapolation of unchecked technology curves—which may be a relevant limit case, but is unlikely to be the probable outcome. These scenarios don’t often acknowledge how hard these complex technologies are to master, and the market and regulatory responses that impede and alter adoption trajectories.

What’s really new in all these discussions about robots and AI? Technological change has been a constant since the original Industrial Revolution. Don’t we always just adapt and move on?

The pace of change is quickening, but folks are also starting to see the convergence of hardware and software into seamless new solutions. At Kiva, we were the first material handling company to create the whole solution—a single integrated complex adaptive hardware and software system coordinating thousands of mobile robots in the storage, movement, and sortation of warehouse inventory in support of order fulfillment processes. This was an eye-opener for the industry, as the efficiency of these deployments would improve year-over-year via better software algorithms and optimizations, new hardware modifications, and changes to workflow processes. A second source of improvements, though, would stem from the “living, breathing” system adapting and tuning itself to better efficiencies. That is the fun and exciting part and the strangely interesting phenomenon that gives rise to these AI speculations. It should be noted, however, that those adaptive improvements were often an order of magnitude smaller than improvements conceived and developed by humans involving rigorously tested new features and algorithms deployed in software updates.

To generalize that observation—concerns around AI and robotics seemingly revolve around not understanding or knowing how these new complex adaptive hardware-software solutions will behave. The increased noise and discussion around the topic simply reflect the fact that we are finally starting to see more fielded examples of such systems. The Internet of Things (IoT) is the same old stuff (hardware) but now with improved, embedded, and connected software that turns this collection into a system capable of useful actions. My car now automatically closes the garage door as I leave the driveway (thanks to a software update earlier this year).

We keep hearing from pessimists about the idea that AI and robots are going to wipe out jobs, and optimists about how technology is going to make us all more productive. And yet neither of those things seems to have happened yet. U.S. unemployment is around 5% (of which probably half is “structural”). And for the past two decades, we haven’t seen any dramatic improvement in U.S. labor productivity. How come the impact of technology—whether positive or negative—doesn’t show up in the economic data?

Though I’m not an economist by any stretch, I would suggest that the impact of technology does show up in the data if you look in the right places. The price of oil was in the $20’s per barrel earlier this year—near the lowest inflation adjusted price ever. This was explained as an oversupply of oil, which I think reflects our technological progress in obtaining the oil that’s been there all along, together with demand-side efficiencies in cars, homes, mowers, and power plants that now use energy more efficiently as well.

When I was a kid growing up, I paid about 50¢ to fill my one-gallon gas can and mow three to four lawns for income ($3.51 during 1981 in 2015 dollars). Last summer I would have been paying about 34¢ inflation adjusted ($2.36 in 2015 dollars), and could probably mow five to six yards before needing a refill. My “mowing business” would be reporting better top and bottom lines today! Maybe corporate profits are where productivity ultimately shows up.

To make another point on productivity, we should be basing our metrics on the number of people employed, rather than unemployed as you simply can’t prove a negative. I suspect that the total output of the U.S. economy over the number of paid hours worked would show that we get more done each year per hour worked (though we probably need to take a fresh look at the GDP numerator too). At the same time, technology also provides a lot of new distractions; binge-watching shows because we can, Facebook, etc., can counterbalance productivity gains.

Is technology going to wipe out jobs? Or is it just going to change them? The McKinsey guys are always telling me not to confuse jobs with “tasks.” They argue you can probably automate half of all tasks, but not that many jobs. Does this sound true to you? And if so, what does it mean?

Jobs versus tasks is a decent framework for thinking about employment and automation. At Kiva, we automated the walking task that consumed about 70% of the pickers’ job time, but we didn’t automate the pickers’ job. When you look around the office environment, you see plenty of repetitive processes that are being automated to improve productivity, thus enabling the knowledgeable worker to focus on the more value-added tasks that actually drive the company forward.

On this point, I always joke that the blame goes all the way back to Gutenberg in 1440 AD, who put thousands of monks out of business when he invented the moveable type printing press, and thus transcribing the Bible and other important works page-by-page by hand was no longer necessary. We know those monks didn’t tweet their grievances, instead, they may have even thrown a party (maybe that’s when they began to focus on beer brewing). And ultimately, civilization benefited from the faster proliferation of printed knowledge.

The fact is that every single new technology, not just industrial waves, change the employment landscape. RF toll tags displaced human collectors standing in smoggy booths. The loom displaced the Luddites. The horseless carriage displaced the blacksmiths. Microsoft Word displaced the traditional secretary. Even the ranks of brave New York City bicycle messengers are more scarce now as email and electronic document signatures have taken over.

In today’s headlines, cab and truck drivers (monks) are high on the speculative list to be sidelined by automated self-driving vehicles (printing presses). While they probably won’t throw a party (after tweeting and litigating their grievances), a few decades from now we’ll all be scratching our heads about the time we used to allow humans to control multi-ton lethal projectiles by hand, in the same way it now seems incredulous that kids once had free range of the back seat without car seats or even seat belts.

Amazon’s experience with Kiva suggests that robots actually make the workplace better, and can make work more meaningful. Do you think that will be true in other settings as well?

Yes—when you automate the mundane portion of a task, the overall job becomes more interesting and meaningful. The human is asked to engage in higher order thinking, creative tasks, and problem solving. That’s true regardless of the setting.

What CAN’T robots do?

Never say never, but today’s robots can’t even pick a t-shirt from a clothing bin, let alone bubble wrap a wine goblet, or fold a pack slip into that sticky window on the outside of an e-commerce box. Those are mechanical tasks that will likely be tackled eventually, but even at that point, the bot will not be smart enough to determine whether a unit is damaged versus sellable, let alone understand the why of the situation.

If I’m a new business school graduate, what should I do to prepare myself for a career in this brave new world? How can anyone train for a job these days when the definition of what employers need seems to change so rapidly? How do I even think about what constitutes an actual “job”?

Don’t think about getting a “job”—think about creating change, moving the needle, pursuing a passion, or changing the game—and develop and use those skills that further those trajectories. Creativity, problem solving, convening, and motivating are skills that won’t be replaced by automation anytime soon.

What can business schools do to prepare students for the new workplace?

Most B schools have started emphasizing the importance of group projects involving teamwork as central to learning the people skills necessary for success in the new workplace. Then layer on some entrepreneurial challenges and context, and you’re moving in the right direction.

What does all this new technology mean for CEOs? How will it force them to think differently about their companies?

CEOs will be those individuals who understand these workforce and marketplace dynamics, and will create companies and solutions that ride these undercurrents as opposed to getting washed away by them. They will place greater importance on finding flexible, adaptable, creative human talent, and provide them with the productivity tools that allow them to automate the mundane and focus on the strategic priorities that grow the business.

Does the “fourth revolution” (if we want to call it that) favor big companies or small companies?

Without even researching the point, I’m going to say smaller companies always have the advantage when things shift or when transformations take place. Each industrial revolution was characterized by new little companies, individuals, and ideas that eventually became big companies and huge industries. Think of Ford competing with hundreds of existing auto startups in the early 1900’s and breaking down the auto patent pooling control of the largest incumbents of the day, or Edison and Tesla/Westinghouse bringing electricity out of tiny labs in an early standards war, or Fairchild Semiconductor becoming a startup that grew too big and too slow, and thus giving rise to “Fair-children” including Intel, AMD, and others resulting in Silicon Valley.

New graduates starting a business career have plenty of challenges: landing a job, finding a place to live, scrambling to make rent—all while trying to plot the next step in their career. Increasingly, it looks like they’ll need to add another worry to the list: competing with robots.

If that sounds like the premise of some screwball sci-fi movie (or reminds you of this Flight of the Conchords video), think again. Over the past several years, a growing chorus of experts, including economists, technologists, and management consultants, have begun warning of widespread job losses in coming decades as advances in artificial intelligence and automation enable machines to take on more and more complex tasks.

– Gartner, an information technology and research advisory firm, estimates a third of jobs will be replaced by software, robots, and smart machines by 2025.

– In a 2013 study, Oxford professors Carl Frey and Michael Osbourne found that machines could replace about 47 percent of our jobs over the next 20 years.

– The McKinsey Global Institute recently concluded that, just by implementing technologies that already exist, global businesses could automate 45 percent of the activities they now pay workers to perform.

Some experts are less gloomy. J.P. Gownder, an analyst with the Boston-based high tech research firm Forrester, estimates that new automation will cause a net loss of “only” 9 million U.S. jobs by 2025. Gownder argues that even as it wipes out some jobs, automation will create new ones, including some entirely new job categories we haven’t even thought of yet. Andrew Moore, dean of the School of Computer Science at Carnegie Mellon University and former head of robotics at Google, says he finds no evidence technology is stealing jobs.

But the list of pessimists includes an imposing roster of science and technology heavyweights. Physicist Stephen Hawking worries about “anthropogenic AI”—robots that might decide to kill us off. He told the BBC recently that AI and robots “could spell the end of the human race.” Elon Musk, head of Tesla and SpaceX, calls artificial intelligence humanity’s “biggest existential threat” and has donated millions to efforts seeking ways to keep AI from turning on its creators.

Steve Wozniak, co-founder of Apple, recently told the Australian Financial Review, “computers are going to take over from humans, no question,” adding, “the future is scary and very bad for people… Eventually computers will think faster than us and they’ll get rid of the slow humans to run companies more efficiently.”

“Eventually computers will think faster than us and they’ll get rid of the slow humans to run companies more efficiently.”

Even former U.S. Deputy Secretary of the Treasury Larry Summers, long an advocate of technology’s unalloyed benefits, has changed his view, writing: “Until a few years ago, I didn’t think this was a very complicated subject. The Luddites were wrong and the believers in technological progress were right. I’m not so completely certain now.”

Luddites, of course, were 19^th century British textile workers who rose up in rebellion against labor-saving production techniques. As that reference implies, the debate about whether technology destroys jobs goes back centuries—and the doomsayers have always been proved wrong.

One of techno-optimists’ favorite examples is agriculture. Two centuries ago, more than 80 percent of the U.S. labor force worked on farms. Today, thanks to advances in mechanization, farmers account for less than 2 percent of the U.S. workforce and we have more food at lower prices than ever.

But Martin Ford, author of the recent book Rise of the Robots: Technology and the Threat of a Jobless Future, argues the technologies driving the current transformation of the economy are different. Improvements in agricultural methods, he points out, weren’t readily transferrable to other sectors of the economy. By contrast, today’s technologies make use of broad-based, general-purpose intelligence. Machines are animated by algorithms that enable them to adapt, to learn, to think.

These brainy new bots are part of a host of other technological changes so sweeping that organizers of this year’s World Economic Forum in Davos, Switzerland proclaimed the dawn of a “Fourth Industrial Revolution,” a brave new world in which “billions of people connected by mobile devices with unprecedented processing power, storage capacity and access to knowledge” are conjoined with “emerging technology breakthroughs in fields such as artificial intelligence, robotics, the Internet of Things, autonomous vehicles, 3-D printing, nanotechnology, biotechnology, materials science, energy storage, and quantum computing.”

In the Davos chronology, the First Industrial Revolution used water and steam to mechanize production; the Second used electric power to create mass production; and the Third used electronics and information and technology to automate production. And now, in the Fourth, we are experiencing a “fusion of technologies that is blurring the lines between the physical, digital, and biological spheres.”

An early milestone in the emergence of thinking machines was May 1997, when IBM’s Deep Blue supercomputer defeated Garry Kasparov, then the world chess champion. Then in 2011, IBM’s supercomputer, Watson, triumphed over Jeopardy superstars Ken Jennings and Brad Rutter.

A year later, Boston-based Rethink Robots rolled out Baxter, a friendly looking robot designed to work alongside humans on the factory floor. Baxter wasn’t a multi-million-dollar super-computer; it was built to human scale and priced at $25,000. Baxter doesn’t have to be programmed—it can learn from watching the movements of people, or by allowing humans to move its flexible arms—and can beat any human at the popular logic game Connect Four. Thanks to recent upgrades, Baxter’s grip is soft enough to cook an egg.

More and better Baxters are coming. Amazon already has thousands of robots sorting items in its fulfillment centers, and envisions replacing delivery workers with a fleet of aerial drones. As of February, there were more than 260,000 robots working in U.S. factories, according to the trade group Robotic Industries Association.

China, which last year became the world’s largest market for industrial robots, means to be at the vanguard of this revolution. Foxconn, China’s largest private employer and a primary supplier to Apple, announced this year that it had installed robots to eliminate 60,000 jobs in a single factory.

Factory jobs aren’t the only jobs at risk. Any kind of office work that involves repetitive tasks such as filling out reports or preparing spreadsheets can be easily replaced with software. Computers are certain to take over many tasks now performed by paralegals and lawyers, especially in big cases where the discovery process can involve millions of documents.

Factory jobs aren’t the only jobs at risk. Any kind of office work that involves repetitive tasks such as filling out reports or preparing spreadsheets can be easily replaced with software.

The Associated Press has shown that computers can generate error-free corporate earnings reports and cover certain types of sporting events. Watson is far more accurate than human doctors at diagnosing lung cancer. A significant share of tasks performed by what are currently high-paying occupations such as financial planners, physicians, and equities traders can be automated with existing technologies.

And will we really need so many taxi drivers when ride-hailing platforms like Uber and Lyft can be paired with self-driving cars piloted by Google, Amazon and Baidu?

What will be left? It isn’t clear. Optimists insist robots will free humans from drudgery and enable us to focus on the types of imaginative tasks in which humans excel. And of course companies will still need engineers to design, program, oversee, and maintain all those computers.

Ford and others fear that, while advances in technology may not eliminate human work completely, they are certain to polarize the labor market, creating higher demand for a smaller number of people with a limited set of professional and technical skills.

How to survive this transformation? The best advice we can offer is to be flexible, keep learning, and prepare for the way you work to change far more rapidly than anything experienced by previous generations.