Technology, Luddism, and ‘Rise of the Robots’

Ever since the Industrial Revolution, the effect of automation on the economy, employment, and society itself has been the subject of heated debate. The claim that rapid technological progress would displace workers and create a surge in permanent unemployment has been advanced by generations of technophobes, often with sound reasoning.

Yet the skeptics have been proven wrong every time so far. Productivity has progressed in tandem with technology, leading to ever-higher standards of living. Moreover, there has been no discernible correlation between technological progress and unemployment. This time, though, might be different.

In his book Rise of the Robots, author Martin Ford eloquently analyses the economic risks we face as technology becomes ever more ubiquitous, affordable and, most importantly, smarter. The overarching concern for Mr. Ford is that with the advent of robotics, artificial intelligence and the egregiously abused “big data” as more mainstream technologies, machines, rather than enhancing the productivity of human workers, might reduce the demand for labor in the long term. And unlike previous times, even white-collar workers may soon face disruption. Indeed, any job described by the author as “predictable” is possibly in danger.

The book begins with a sweeping overview of the current state of automation. It starts with warehouse robots which, in the confident conclusion of their makers, will eventually be able to move one box every second, compared to one box every six seconds for a human worker, taking advantage of advanced computer vision capabilities. The case study serves as an exposition of Mr. Ford’s theme: a highly repetitive, predictable job constitutes a perfect target for automation by machines which are able to work continuously, will never “suffer a back injury — and it will certainly never file a worker’s compensation claim”. Nor will they, one might add, unionize.

The narrative continues through lightweight manufacturing robots, the service and health care sectors, self-driving cars and language processing technology. In the service sector, automation in fast food restaurants and retail is discussed. Namely, replacing fast food employees flipping burgers or designing retail stores with no human cashiers whatsoever (à la Amazon Go). The elimination of the demand for labor in these industries would be an unwelcome development, as they provide a stable income to less educated members of the workforce, many of whom nowadays tend to support families.

So far the most politically loaded profession potentially threatened by technology has been that of truck drivers. According to a widely reported (though disputed) statistic, ‘truck driver’ is the most common occupation in the majority of U.S. states, with about 2.8 million workers choosing it as their primary occupation. The recent trend towards self-driving cars, which have achieved outstanding precision and control (despite widely publicized mishaps), have left economists worried about their impact on the workforce.

The usual response of tech-enthusiasts is that the disruption of education caused by the numerous massive open online course (MOOC) platforms will allow workers to acquire new skills and become more competitive in the fast-changing job market. Yet this will remain wishful thinking, at least for the foreseeable future. As Mr. Ford expounds in a chapter dedicated to the MOOC phenomenon, MOOC’s today are firmly focused on providing additional skills to people who are highly educated and self-motivated to begin with. And even among them, the completion rates of online courses are appallingly low. While MOOC’s might end up making a difference to the less educated, they still have a long way to go.

Yet Mr. Ford is at his most persuasive when he argues that in this new Industrial Revolution, unlike the previous ones, the jobs in danger are not only the blue-collar ones. Even many white-collar jobs for the well-educated are at risk. With respect to those jobs, the rise of artificial intelligence plays an even greater part than in the blue-collar jobs, and one must, perforce, spend some time understanding its implications.

Developing intelligent machines is hard. As computers have grown exponentially faster (following Moore’s law) ever since their debut on the world stage, the AI pioneers may be forgiven for believing that human-level intelligence was just around the corner. No less august a figure than John McCarthy (the creator of Lisp) mused on creating a fully intelligent machine in a decade. The year was 1963. Yet the complexity of humans’ cognitive processes inevitably got in the way. Indeed, this bleak history of under-delivery by the AI community might give pause to anyone certain of the imminence of superintelligent machines.

What makes the recent advances so groundbreaking? After all, Deep Blue, the machine that beat Garry Kasparov in chess, seemed a major breakthrough, but it hasn’t led to monumental advances in machine intelligence. What makes the current AI surge exciting is the fact that, for the first time on substantial problems, AI programs have been able to learn from data, with no presuppositions or strategies explicitly encoded by programmers. This is in sharp contrast with previous techniques in which search trees and problem-specific evaluation functions were used. Just compare the Deep Blue paper to Google’s AlphaGo, which recently beat Lee Sedol, the world champion in the fiendishly complex game of Go [correction: Lee Sedol is not the current world champion].

The applications of these so-called deep learning techniques span a large number of problem spaces, including image recognition and classification, game playing, language processing, medical diagnostics, robot control and so forth. One reason why these techniques seem so promising is that they generalize very well. As Mr. Ford helpfully points out, the DeepMind team at Google has trained a neural network to play several Atari games using essentially the same architecture and algorithms. Some would argue that this is a big step in the direction of general intelligence.

The implications are profound. Advanced language-processing capabilities may conceivably obviate the need for translators, as well as writers of more matter-of-fact news stories, not to mention paralegals and lawyers dealing with more mundane tasks. One could conceive of radiologists and other diagnostic physicians being replaced. In general, any job that is repetitive and depends mostly on pattern recognition could be a target for automation, no matter the level of education currently required.

Apart from the obvious possibility of creating mass unemployment, the author warns of a more subtle and insidious economic consequence of automation — a long-term collapse of consumer demand. He illustrates the danger by the famous anecdote:

There is an often-told story about Henry Ford II and Walter Reuther, the legendary head of the United Auto Workers union, jointly touring a recently automated car manufacturing plant. The Ford Motor Company CEO taunts Reuther by asking, “Walter, how are you going to get these robots to pay union dues?” Reuther comes right back at Ford, asking, “Henry, how are you going to get them to buy your cars?”

In light of these maladies, the author’s prescription is one that has been consistently gaining traction — a universal basic income. The premise here is not the oft-invoked Marxist fantasy that people, free from want, can be expected to dedicate themselves to loftier pursuits. It is simply that many of the jobs we now take for granted will end up obliterated, leaving the UBI as the only viable option.

The book, to be sure, is not without flaws. It is not as dispassionate as one would hope. There is much Piketty-esque rhetoric on the relation of labor and capital. We are advised that:

The fact that both Apple and Microsoft were founded in the mid-1970s — a period when the top tax bracket stood at 70 percent — offers pretty good evidence that entrepreneurs don’t spend much time worrying about tax rates,

confidently generalizing based on a statistically significant sample of two.

A particularly confusing aspect are the remarks on Moore’s law. For a large part of the book, the author seems to imply that Moore’s law is still its old self, doubling computing power every 18 months, even though the consensus among technologists is that it has slowed down substantially. Mr. Ford acknowledges as much explicitly, even providing an excellent S-curve explanation of technological progress.

When discussing the effect of automation on white-collar jobs, the author points to several systems that seem to have no particular relevance to the economic predictions that he makes. There is Eureqa, touted by its creators as something of an ‘AI scientist’, capable of discovering natural laws independently. In reality it uses genetic programming, widely considered a stagnating field, and certainly rather far removed from any semblance of actual intelligence. Also touched upon are efforts in automated musical composition and art. Though indisputably impressive, these paint a picture of AI far more advanced than it is in reality.

The math on the universal basic income seems somewhat shaky. Presumably the basic income is more than glorified social security, allowing people to live decent, though not particularly comfortable lives, and encouraging “robust consumption”. Yet in attempting to prove the budgetary feasibility of such a scheme, Mr. Ford proposes a figure of $10,000 per year, even less than what the average social security recipient gets (around $1,300 per month). Altogether, it seems to me that the author significantly underestimates the cost of the basic income.

Yet even with these reservations, ‘Rise of the Robots’ is a refreshing and imaginative book (despite its gloomy predictions), and one of the first to offer a comprehensive look at the ways technological progress in the age of smart machines is likely to reflect on the economy and society. It does away with dystopian hysteria, focusing instead on economic facts and trends, even offering practical advice. And it reminds us, in an ever-changing world, not to blindly embrace technological progress without thinking through the consequences. Kudos to Mr. Ford for that. And I certainly look forward to hearing what else he has to say and write on these issues.

Special thanks to Ivan Mandura and Ognjen Stipetic for reviewing drafts.

Building https://www.quicknews.ai/. Ex-Amazon Engineer, on AWS and Alexa. Vancouver, BC.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store