Discourse on futurology often takes on quixotic airs; within the context of seemingly exponential technological growth, it becomes easy to view future technology as an unyielding force for engineering social and industrial change. With the greater productive efficiency that technology brings, it is said that we can all get more things at lesser cost. What tends to be omitted from the most idealistic of these projections are the structural reverberations their realisation may send across the relevant sectors’ attendant labor markets. When poverty and unemployment do enter the equation, the problems they represent are framed as problems of distribution, themselves something to be remedied by technological and social innovation.
For every job destroyed by increases in efficiency, it is often assumed that a new, better job will appear to take its place. After all, if humans don’t need to do menial tasks, they can be free to pursue other, more creative, more “fulfilling” goals. Fearing that this may not be the case seems a sure way to be called a luddite, and indeed economists even use the term “luddite fallacy” to denigrate the notion that unemployment in specific job markets affected by technological growth (technological unemployment) can lead to overall unemployment in the labor market as a whole (structural unemployment).
The Industrial Revolution provides a textbook example of massive technological advancement that caused widespread shifts in employment from some fields to others, leaving overall employment rates relatively intact. By creating machines that augmented the physical capabilities of human labor, fewer people were required to do the same jobs, and the increased production dramatically brought down prices for goods, thus allowing people to purchase more items, which in turn drove demand up, created more jobs, etc. The service industry, as well, was made possible by an expanded middle class that could spend money on things beyond food and shelter. Today, a new revolution looms. Brought upon by breakthroughs in artificial intelligence (AI), big data, and robotics, computers are rapidly improving their cognitive capacities and are threatening to transform the labor market as profoundly as the Industrial Revolution did. Now, are there any grounds to heed the luddites’ worries this time around?
This possibility is worth considering for two principle reasons. First, because this revolution deals with the cognition of machines, there will be permanent losses in low-skill labor options, precipitating the possibility of entry level jobs requiring qualifications that are unachievable for many people. Secondly, the global capitalist economic system currently in place would concentrate wealth and power in the hands of those least inclined to distribute the benefits of technological growth: the owners of the technology itself.
The first of these problems would obviously be directly allayed by increasing the average intelligence of the labor pool to match the requirements of this new generation of jobs. Assuming that better schooling and childcare isn’t enough to do this, there are two remaining options, enabled by the futurist context of the problem at hand: eugenics, and artificial cognitive improvement. The former of these is marred by the obvious, concomitant moral considerations, but finding ways to improve the way we think through technology seems a poetic way to address technologically-caused unemployment. The most prominent futurological avenue to carry this out would be the brain-implanted, physical “add-ons” of transhumanism. Unfortunately, the very economic problems that this option is meant to address are what make it so hard to implement. Complicated, cutting-edge medical procedures of the kind that would be necessary to actually implement transhuman modifications, were they to ever be realised in the first place, would almost certainly be restrictively expensive. As such, it would seem likely to further exacerbate inequality, as the rich and powerful within a transhumanist reality would be able to directly leverage economic inequality into intellectual and genetic inequality.
Further complicating the matter is that the technology that could potentially cause permanent structural unemployment may come before the technology or legalization of transhumanism. Indeed, self driving cars, automated fast food service, and robotic manufacturing are all on the brink of automating millions of jobs, while the genetic basis underlying intelligence is no where close to being understood. Human intelligence itself, to be frank, is poorly understood by both cognitive and computer scientists, despite the fact that both fields are in their “golden ages”, while, as previously mentioned, AI research has been making particularly fast progress in creating intelligent systems.
The big leaps AI research has made, however, are generally still within the frameworks of specific tasks, not the generalized – and flexible – intelligence characteristic of human beings. Contemporary machine learning is most predominantly implemented via methods within the domain of “supervised learning”– a set of techniques in which machines learning a given task are shown examples of successful completions of the task, and given feedback (in the form of “error signals”) to correct their performance. In human learning, things like the laws of physics and object segmentation are innately understood through exploration and exposure in the real world; you do not need to be Isaac Newton to heuristically infer certain physical realities from viewing an apple fall from a tree. Because research is so much more advanced regarding the aforementioned, narrower forms of intelligence, applications that rely on intelligent behavior for specific tasks are going to come much earlier than those that rely on general intelligence. Since jobs by and large require limited sets of behavior, and transhuman modifications would require holistic understandings of the human mind – not to mention ways to deal with the complex psychological impact of actually being implemented – it is almost certain that large swathes of the job market will be automated before there is any significant adoption of transhuman modifications.
What we still have not established, however, is exactly why this upcoming revolution would cause unemployment that is structural in nature, rather than just technological, as in the Industrial Revolution. The short-term reason is that we can’t think about automation in a vacuum: it’s effects are being filtered through the modern capitalist economy. Blue collar labor in North America is already rapidly disappearing, and unemployment rates for those without a college education are significantly higher than average.
While this in itself is not a terrible thing – after all, college education should provide better working opportunities – it is rather concerning when one considers the prohibitive and ballooning costs of college education. With no job market, and no money to move into a different one, blue collar workers replaced by automation will be faced with large-scale unemployment and poverty. This will lead to ever growing inequality as the wages that would have once been paid to workers are now turned into investments for the hardware and software created by the smaller technocratic class.
Let us assume for a moment that AI-based automation goes beyond self-driving taxis, and takes a larger role in the economy, eliminating many skilled jobs, as well as unskilled ones. In fact, it seems likely that only when AI is advanced enough to automate these skilled jobs, will the technology necessary for transhuman modifications be available. In this environment, there may be a class of “unemployables” who lack the necessary abilities to maintain a job that has not been automated, and who would require transhuman modification to obtain those abilities. Unless the operations are free, these are the same people that would not be able to afford those modifications to function. Furthermore, even if they are free, there is something morally problematic about forcing people to undergo a radical operation just to be functional in society. In either case, the possibility of transhuman modifications would create a class difference between those that attain them and those who do not. The ethical questions surrounding transhumanism and automation would become as important as the technological ones, and careful consideration in how these technologies are implemented would be needed, should we have any hope of resolving the immense attendant power imbalances.