We need a New Deal to address the economic risks of automation

Treasury Secretary Steven Mnuchin has a surprisingly naive take on the issue of AI and robots replacing human jobs.

Mnuchin said that human obsolescence is “not even on our radar screen… 50-100 more years away” — and that he’s “not worried at all.” For those of us in the tech industry, it’s clear his timescale is inaccurate and that, if the government is complacent about AI, the country is setting itself up for an economic shock.

There’s been downward pressure on jobs since the Industrial Revolution due to leaps in productivity brought about by human ingenuity and lucky discoveries. This has accelerated since the 1980s with the mass adoption of computers, but the market has more or less kept up, creating new openings to fill the eradicated ones, albeit not in the same places (coastal cities have gained, Rust Belt areas have lost out).

However, we have a tsunami on the horizon: automation using AI. It will place intense downward pressure on employment, and threatens to catch a generation (really, three generations) off guard, with unemployment levels higher than the Great Depression.

The automation tsunami

Automation is going to leave millions of Americans un- or under-employed not in “50-100 more years,” but within the next decade. As someone very familiar with one of the most-affected sectors, the customer service industry — I invented web chat technology for customer service in the late 1990s, and have been in the space for more than 20 years — I can already see the change happening, and the robots appearing, today.

From our data on large banks, telcos, retailers and so on, we can see that approximately 40-50 percent of tasks performed in customer care fall into a category that is highly susceptible to automation: routine processes like updating payment cards, addresses, passwords and other basic processes. More than 3 million people in the U.S. alone are employed in call centers, and it’s clear to me that the Fourth Industrial Revolution will displace a substantial number of U.S. workers in a far shorter time period than what Mnuchin is positing.

Franklin Delano Roosevelt. Photo courtesy Flickr/FDR Presidential Library

A new New Deal

We need a new New Deal to tackle the consequences of automation. It should be big, aggressive and, unlike the first one under FDR, it should be preemptive. Without some kind of counterbalancing action, we’ll see tens of millions of workers stranded, with curtailed employment prospects. This will kick off a hereditary shockwave of economic hardship that could be felt for generations.

We know the jobs that will be affected first, in addition to the 3+ million customer care agents I mentioned above, and they are very common ones that employ large swathes of the population: truck and taxi drivers, cashiers, security guards, retail clerks and a number of others. These people, who will suffer the most, are also the least aware of new advancements in AI and automation (in fact, they are the least likely to be tech savvy at all). Not only this, but they are highly likely to be in lower income brackets without the luxury of time to re-train while holding down their current job(s), and without the savings to invest in re-education.

More alarming still, our own research shows that 88 percent of Americans reported not being worried about losing their jobs even after being shown research from Oxford University predicting that 47 percent of U.S. jobs are vulnerable to automation. So we have potentially millions of unemployed Americans on our hands in the next two decades, with very few of them currently doing anything to prepare for a new occupation.

3D render of a robot trying to solve a wooden cube puzzle

What’s the answer?

A number of people in tech have begun to offer solutions to the looming automation tsunami. Bill Gates suggested we tax the robots in order to slow the progress of automation and re-direct money toward human-service jobs that require a level of empathy and compassion that artificial intelligence cannot yet offer. Massachusetts also floated an idea of taxing self-driving cars.

The problem with taxation is that it is a deterrent to deploying the technology and reaping the productivity gains, while only indirectly helping those who are sidelined by it. It could also be a nightmare to measure, and actually collect that tax.

Others, like Elon Musk, have suggested that a universal basic income is our best way forward to ensure even those who are left with no options for employment are still able to reasonably take care of themselves. Y Combinator also ran a basic income experiment in Oakland.

I’m skeptical of this approach too, although it’s certainly true that social safety nets will need to be strengthened in the face of the shift to automation. UBI reduces the incentive to work, and risks stranding millions of people in a subsistence living trap, able to just about get by, but cut off from the opportunity for upward mobility, as this essay details well.

Labor, Management

People and platforms

The first step should be to build up flood defenses against the tsunami. We already know which industries are most vulnerable. Focusing on those, the new New Deal should fund education and retraining programs that provide an opportunity for at-risk employees to learn new skills, geared toward those industries that will be around longer-term. The government could partner with major employers to advertise these training and education programs, and make sure staff are aware of them, as the number of open positions at those employers declines due to automation.

While initiatives like this almost always tend to over- or under-shoot, it will at least soften the pain of  impact. Think of it as a stimulus program that incentivizes those individuals in need of job retraining and those organizations that proactively create new jobs. Past programs like Roosevelt’s New Deal and 2009’s ARRA mitigated the worst of the impact of economic turbulence.

My second suggestion is an area where technologists can help. Create software platforms that rely upon human input and labor in areas where AI is less applicable: things like creativity, empathy and other uniquely human capabilities. Government could help with this. For example, it could offer co-investment to startups or companies that create platforms that use human labor to do uniquely human activities. (Usually, government investment in tech goes poorly, and inflates bubbles, but in this case, a “pre-bubble” that anticipates the great automation shift would actually soften the landing.)

We have seen that mass-production in industry has led to a re-evaluation of pre-industrial techniques — artisanal goods, à la Etsy, grew in prominence and value — as people start to cherish the idiosyncratic and the hand-made. The same goes for agriculture, where organic farming and even home farming (in Brooklyn!) made a comeback as an alternative to mass-scale industrial farming. In both cases, tech platforms have helped make those older forms of production economically realistic by reducing distribution and access costs, even though the base cost of goods was higher.

I believe we will need platforms that take advantage of the glut of human labor that the great automation shift will create, and, specifically, we will need platforms that focus on tasks where AI has trouble, and where humans have the upper hand, such as design, creativity, empathy and judgment. Instead of sticking its head in the sand and pretending everything is fine, the government and Steven Mnuchin should focus in more detail on the disruption that’s about to happen, and build a plan to address it.