Big data + smart machines = unemployment?

As more of our lives become datafied, we are making leaps in artificial intelligence to make sense of all the data that we generate. Google, Facebook, LinkedIn, and other tech companies are making AI related acquisitions and hires. In a TED interview with Charlie Rose a few weeks ago, Larry Page gave a lot of airtime to Google’s future in deep learning and the enormous potential of applying AI to our daily lives.

As consumers, big data is great for us. It’s supposed to get us better products and services based on what we want and like. As a society, we’re supposed to want technological progress because it creates jobs and wealth. But the combination of big data and smart machines – resulting in automation of tasks we thought could not be broken down into algorithms and rules – could displace a large percentage of workers in the near future.

A study by Oxford University estimates that AI could take over nearly half of all jobs in the United States in the next two decades. If we compare that to some of the biggest technological shocks history has seen – and if we believe in Andrew McAfee and Erik Brynjolfsson’s predictions about the “second machine age” completely reinventing our economy – then a displacement of labor at that scale seems possible. We’ve seen massive shifts from agriculture to manufacturing to services before.

In the worst case scenario, the upcoming wave of technological innovation will generate fewer jobs than it displaces, accrue wealth to machine owners, and worsen our inequality problem. I believe in a happier outcome and am excited about what’s next in big data and AI. Smart machines will help us make incredible leaps in productivity.

But there’s also a darker side to the story that I wasn’t thinking much about until I ran across the Oxford study. There are always winners and losers, and technological unemployment will be a reality in the near term. We’ll have to endure deep, structural adjustments to a new economy that will be painful even with good policies. If you’re depressed by this post, here’s the upside: we can say with 99% certainty that telemarketing as a job will no longer exist in 20 years (see chart).

Want to keep the conversation going? Weigh in on this article and others at the HBS Digital Innovation and Transformation blog at: //