Ryan Abbott, a law professor at Surrey University, presented a convincing case for such a tax– or, more precisely, for removing the tax incentives that favor automation over human labor. Many business decisions to automate processes, argued Abbott, are driven by these tax advantages, not because the robots are more productive. If automation is more efficient, suggested Abbott, let that be the reason businesses decide to use it – not some tax break.
The counter-arguments from Ryan Avent, economics columnist at the Economist, were that taxes (or ending tax breaks) would slow innovation, and, in any case, there is little economic evidence that robots are taking over jobs. While these points appear to be less convincing, a robot tax alone won’t solve the lack of good jobs. But removing the financial incentives that favor automation over humans will at least create a level playing field.
Even more erroneous is the argument that a robot tax is a slippery slope: are they going to tax my Roomba or smart toaster next? Somehow, we manage to tax labor, and yet you can still mow your lawn and clean your house without paying the government.
Alas, the Emtech Next attendees disagreed, with 70% voting against the robot tax after hearing the debate.
One of the recurring themes at the conference was that businesses could better exploit automation and AI not just for their own efficiency, but to improve productivity and grow the economy as a whole. How? A pair of engineers, Meera Sampath at the State University of New York and Pramod P. Khargonekar at the University of California, Irvine, presented their plan for “socially responsible automation,” which starts with getting technologists to think harder about how their creations will actually be used and how those uses can benefit workers and society.
A shout-out to workers: Too often in these discussions of how automation and AI are affecting jobs, the voices of workers themselves are absent. MIT’s Thomas Kochan, speaking at the conference, at least provided a reminder to listen to such views, even in the early stages of product designs, and to involve them more in automation decisions. And, he argued, companies need to take the time to give their existing workforce the skills and training necessary to integrate them with changes in automation better.
Let’s admit it, every time we hear the word “co-bot” we cringe. Yes, there have been remarkable advances in robotics over the last decade that allow these machines to more safely and comfortably work alongside people and do more human-like tasks. And, yes, we know the promise is that, by taking over mundane tasks, these robots will free people up to do more interesting and, hopefully, productive ones.
But that’s a business decision that too often companies are not taking; instead, many are simply replacing their workers. If robots do 20% of the tasks that a worker was doing, then you need 20% fewer people to get the job done.
It’s become increasingly clear to economists that this one reason we’re facing a crisis: wages are flat, and job opportunities are limited for many workers.
MIT economist Daron Acemoglu blames this on what he calls “so-so” automation and technologies. Advances like automation should be a boon to productivity, but productivity growth has been sluggish for more than a decade. That, says Acemoglu, is because too often companies are automating jobs even when the machines are not more productive, because of the tax mentioned above distortions and a general enthusiasm for robots. So, you have a double whammy; not only are robots replacing workers, they’re not particularly adept at growing the economy.
The way out is to create new, productive tasks for the workers replaced by the automation. (That’s what happened in the past). And that’s where AI could be useful. Examples are not that hard to imagine. For instance, if you free up healthcare workers, such as radiologists and nurses, from routine tasks, they could use AI systems to collect and analyze far more patient data, expanding their capabilities and giving them new ways to advise and treat patients. Acemoglu cites similar examples existing in education and manufacturing.
But, and this is key, Acemoglu warns that this won’t necessarily happen on its own. You can’t leave this up to the markets or the technologists. We need to pursue this goal deliberately.