The future of automation and the automation of the future

Chaplin’s 1936 Modern Times, by now a classic of silent cinema, offers an inside glimpse of the automation of industrial production in the first part of the 20th Century. Our little tramp has somehow found a job in a factory, which is in the middle of an unspecified city, and spends his time doing the same minute but mechanical task over and over.  As he is part of a much larger assembly line, there is no room for error – which, needless to say, is bound to happen, this being a Chaplin film after all.  In any event, we can see the wonders of electro-mechanical automation accompanied by the dramatic reduction in skills workers needed to have back then to get a job in a factory.

32 years later, Kubrick’s futuristic 2001: A Space Odyssey depicted a very different kind of automation. Our main star here is  HAL (Heuristically programmed Algorithmic computer),  a powerful and brilliant computer that calls most if not all the shots of the space trip it is undertaking alongside real human beings. The role of the latter is pretty much reduced to checking that HAL is not malfunctioning. In on the first scenes aboard the space ship, we see one of the astronauts playing chess with HAL – perhaps a prelude to the 1997 match between the then chess champion of the world Gary Kasparov and  IBM’s Deep Blue computer which was won by the latter.  Here we see the splendor of digital automation and its potential cognitive impact on the people who happen to work in such environments.

History can indeed teach us quite a few things, especially in an era where “innovation” and “disruption” seem to be its clarion call. Automation is not new and has been around since the 18th century at least. It also appears to be a distinct trait of capitalist development.

In his book on the subject, Carr argues that what is really different today is that automation is mostly digital. We have already left the mechanical age, and now computers and algorithms are pervasive – and invasive. We are, however “enamored of automation” (pg. 17). Why? Because of its great and most obvious benefits. We, however, do not stop for a moment to think about its impact on the labor market and on our personal lives. We thus need to make some “ethical choices” (pg. 18) to counterbalance the forceful and almost “automatic” automation push.

An early example of digital automation can be found in the airplane industry. We all have heard about the “automatic pilot”  feature in airplanes which in essence lets a sophisticated computer system, much a la HAL but without the neutral, inflexible voice, take full charge of the plane. The issue here is that “automation degrades pilot performance” (pg. 56) as pilots now require a different skill set to be able to manage the computer platforms now installed in most airplanes. The same goes for expert systems, electronic medical records, maps and GPS, architectural design and many other white collar jobs which are being automated nowadays.

Carr’s main point is that automation has a real cognitive impact on human beings which can assume two different forms: automation complacency and automation bias (pg. 67). The first gives us a false sense of security as we tend to believe everything is going to work correctly.  However, if something happens to fail we might be in big trouble as we do not know really know how to fix it.  Automation bias is when we assume that the information we are getting from computers and digital platforms is always accurate, even if it is, in fact, misleading or wrong. Carr argues that humans are susceptible to these two cognitive impairments and thus allow automation to create a gap between themselves and the knowledge they have. As a result, “…automation distances us from our work, … it gets between us and the world, it erases the artistry from our lives” (pg. 85)

How do we address this?  How can we close this gap and avoid such a loss?

The answer, Carr says, is by having human-centered automation. That is, we need to design automation processes which put people, who are potential beneficiaries, at the center of the process, thus avoiding a purely technological approach to automation. And this is not really difficult to do as Ergonomics for example shows.  Of particular interest here is adaptive innovation that fosters the interaction between algorithms and software, and the end user. There is a division of labor between the two that can be adjusted as they interact with each other. In this approach, software supplements “human judgment” (pg. 166) and does not try to supplant it.

So why isn’t this happening today? The blame is undoubtedly not on software developers and geeks who tend to emphasize technology. After all, they are just responding to the demands of businesses and the short turn pursuit to maximize profits, increase productivity and reduce costs. In this context, we need to understand all the socio-economic, political and ethical underpinnings that drive software development to avoid being manipulated (pg. 208).

As technology becomes even more invisible, we need to dive deeper to fully grasp what is really going on. The motivation to undertake such a task, in the end, has to come from our own desire to define the real meaning of being a human being, concludes Carr.

This, however, might not be the only or best response to the issue. If automation is a social process, it is then difficult to find a solution in our inner desires. Another more plausible option could be to democratize automation processes and allow stakeholders and beneficiaries be part and parcel of the design and implementation of new digital solutions that will make our lives better without us becoming dummies in the process. The question here is the feasibility of such an option in our current socio-economic and political context.

Carr, Nicholas. 2014. The Glass Cage: Automation and Us. New York: W.W. Norton & Co. ISBN: 978-0-393-24076-4.

Cheers, Raúl