Make Jobs Great Again

Though the future of work is a very big issue, in the 2016 US presidential campaign it was explored in a very small way.

By Lisa Conn

The next President will face serious questions about the future of work in a technology-driven and increasingly automated economy. Advances in artificial intelligence, machine learning, and robotics are rendering many jobs obsolete, sparking an ongoing national debate. However, the topic has not figured prominently in the election discourse this year; instead, the most discussed issues have been Immigration, Foreign Policy/National Security, and Guns. In the recent debates, for example, none of the moderators asked the candidates how they plan to respond to the automation revolution.

Here at the Electome, a project of MIT Media Lab’s Laboratory for Social Machines, we use advanced computer science techniques to understand how people on social media talk about the election. With access to the full output of Twitter, our algorithms classify all election-related tweets — roughly half a million per day — by topic, candidate, candidate supporters, gender, state, and civility. Using three of the issues we track — Jobs/Employment, Trade, and Income Inequality — we analyzed the future-of-work discussion, which generally accounted for less than 5% of the total policy-related tweets since January 2016.

Two Opposing Visions

In late June, however, these issues spiked briefly on Twitter when Clinton and Trump gave competing speeches about the economy on the same day.

Clinton spoke at a tech workforce training facility in Denver, Colorado, and Trump at a scrap metal factory in Pennsylvania — both battleground states. Clinton sketched an economic future based on continued innovation. She outlined a technology plan that expands STEM education, defers student loans for entrepreneurs, and reduces regulation for tech start-ups.

While the plan was received positively by technologists, others noted that it didn’t offer much to today’s displaced workers — who figure prominently among Trump’s core supporters.

Meanwhile in Pennsylvania, Trump called for a revival of more traditional industries. For instance, he critiqued NAFTA for sending US manufacturing jobs overseas.

Trump did not address the impact of automation on traditional factory jobs, which are expected to decline 22% by 2025.

First presidential debate

The second spike for this topic occurred on September 26th, the evening of the first presidential debate. Moderator Lester Holt asked the candidates about the industries that have left the U.S. for cheaper labor overseas.

Clinton mentioned “the jobs of the future.…That means jobs in infrastructure, in advanced manufacturing, in innovation and technology, clean renewable energy, and small business.” By contrast, Trump talked mainly about the jobs that the country has lost — not through technological advancement, but outsourcing. “Our jobs are fleeing the country. They’re going to Mexico…thousands of jobs leaving Michigan, leaving Ohio. They’re all leaving.”

The stark divide between the candidates’ approaches did not go unnoticed on Twitter. While many Trump supporters cheered his emphasis on ending outsourcing, other tweeters highlighted the automation problem.

Beyond Tuesday

Though the future of work is clearly a very big issue for the country, in this campaign it was explored in a very small way. Where is the American workforce headed? The 2016 election didn’t answer this question, but there are lots of clues and data-points. For example, Boston Consulting Group estimates that approximately 25% of manufacturing functions will be automated by 2025, while demand for high-skilled workers to operate advanced machinery is expected to grow.

One important, non-automated job is up for grabs on Tuesday, and whoever gets it will almost certainly find this question looming large.

Lisa Conn, an MBA student at MIT Sloan School of Management, has been working on the Electome project at the Laboratory for Social Machines, part of the MIT Media Lab, since last fall. This piece was written using analytics from the Electome dashboard.