AI: The Zombie Apocalypse is Not Upon Us, or is it?

A story on today’s NPR Morning Edition raises the specter of the Zombie Apocalypse. So, too, do several Hollywood movies. I am not speaking about real Zombies. Instead I am speaking about our innate fear of and attraction to the liminal space between human consciousness and AI. Chappie, Ex Machina, Avengers: Age of Ultron, Transcendence all play with this liminal space. Are they dead or are they living? And what do they want to do with us?

One of the motifs repeated in many of these movies is the idea that AI will cost us our jobs or, in any case, deprive us of our need to work. And so we are fed images of police stations with humans, feet on desks, coffee mugs in hand, relaxing while AI does the real crime fighting. When needed for real action, the humans are completely unprepared. A similar picture is painted in Wall-E, where, after generations of being served by AI, humans simply vegetate.

And, yet, clearly there is a real issue at stake here. Which brings us back to David Kastenbaum’s report: “Experts Debate: Will Computers Edge People out of Entire Careers?” David’s report features two MIT professors: Andrew Paul Macafee, co-director of the MIT initiative on the Digital Economy and author of Race Against the Machine (2012), and David Autor, Associate Head of the Economics Department and editor of Studies of Labor Market Mediation (2009). Professor Kastenbaum warns that as machines get smarter, cheaper, and more skilled they will begin to supplant human labour, which, obviously, is already taking place. Professor Autor says not, but he gives an inapposite example for why not: folding laundry. The example is inapposite not only because of Alex Cartier, the high school student featured on Youtube who built a laundry folding machine. (There are hours of laundry folding machine footage on Youtube, Alex’s and others.) It is also disturbing to think that this might be the kind of work left to us once AI does take over. Now that’s scary.

I was truly saddened that David Kastenbaum failed to invoke the first expert on this subject: Georg Wilhelm Friedrich Hegel. It was Hegel (and not Adam Smith) who first pointed out why mechanization might not be so bad a thing after all. Hegel showed that as our tasks become ever more specialized, as we break them down into ever smaller and simpler units, we will eventually be able to hand these tasks off to machines:

At the same time, this abstraction of one man’s skill and means of production from another’s completes and makes necessary everywhere the dependence of men on one another and their reciprocal relation in the satisfaction of their other needs. Further, the abstraction of one man’s production from another’s makes work morf and more mechanical, until finally man is able to step aside and install machines in his place (Philosophy of Right §198).

Cool, huh. Only problem? Didn’t happen. And this gets to the heart of why Professor Macafee may be mistaken. Sure, today we produce more stuff, for a lower cost, for more people than ever before in history — largely because of technological innovation. But, then why are we not able to “step aside and install machines” in our place?

The answer lies in a misreading of why we work. The arm-chair philosopher answer to this question is: duh, we work in order to produce things that we need or desire, which other people can purchase because they work in order to produce things that we need or desire, which . . . However, cultural anthropologists such as Nurit Bird-David, now at Israel Institute of Technology, point out that most human communities work only so much as they have to and invest the remainder of their time on what would today be called community-building activities (see The Original Affluent Society). Once they had enough, they stopped. It is really only within the last five hundred years that human beings have felt compelled to arrange their lives so that, when they have produced enough, they find something else to produce. Assuming that hominids have been around for roughly 2.5M years, five hundred years is a negligible span. For roughly 98% of their existence on the planet, hominids have preferred not to work.

Were our society based upon satisfying human needs or fulfilling their desires, we too might anticipate an end to work, just as Professor Macafee fears. But let us suppose that we work not in order to satisfy human needs or desires, but rather in order to create sufficient efficiencies for people who don’t work. (And I am not thinking here about welfare moms who, let me tell you, work plenty.) I am thinking here about the top of the income hierarchy who live off the efficiencies produced by those beneath them. Yes, we could have a machine fold our laundry; but for two problems. First, there are some things that even machines won’t do. Wages and benefits may have been driven down so far that it is actually more efficient and less costly to hire a human being at a marginal wage to fold and press my shirts. Second, unless I am simply going to hand out money so consumers can purchase what I produce, I need to have consumers do something, anything — bagging, check-out clerk, soda jerk, retrieve my golf balls, make and serve me my meals, care for my lawn, my children, cater my parties — anything to maintain the dependent relationship between the 99% and the 1%. We work not for lack of plenty of technology that could do our work for us. We work because that is how social relations are mediated in our society. Or, put differently, we work in order that our efficiencies can be sent up the income hierarchy to those who don’t have to work.

So, not to worry. There will be plenty of work in the future even if AI takes over. Who knows, perhaps some day you and I will be lucky enough to work maintaining some Android at a standard of living to which she has grown accustomed. Perhaps we already are.

(Joseph Lough teaches Economic History and Theory at the University of California, Berkeley.)