The debate about whether or not robots are stealing human jobs is a perennial one. One side of the debate argues that the automation heralded by 21st-century technologies is leading to higher levels of unemployment—what the economist John Maynard Keynes referred to as “technological unemployment”. He described it as jobs being lost to automation at a rate outpacing the invention of new jobs.
The opposing view is that the threat of technological unemployment is mitigated by new jobs. These have repeatedly replaced those made redundant by successive industrial revolutions. Contemporary proponents of this view argue that the overall effect of the current artificial intelligence (AI) revolution is that some jobs are being lost while others are undergoing modification, but quite significantly, new jobs are also being invented.
It’s hard to disagree with the latter view, given the emergence of previously unimagined jobs. For example, when YouTube was invented 20 years ago, it’s unlikely that its makers imagined that their video-sharing platform would one day be raking in millions in revenue for social media content creators, an unheard-of profession in those days.
Nevertheless, we still find ourselves caught up in an ongoing debate about technological unemployment. Despite the opportunities heralded by the cutting-edge technologies of our time, sustaining livelihoods somehow seems to be getting harder. This intuition is, in fact, backed by a report from the ILO. At the beginning of this year, it projected a worsening global unemployment rate, whilst reporting that disposable incomes are in decline alongside a general drop in the standard of living.
So, what’s going on? Why haven’t the technologies of our era yielded better socio-economic outcomes for society?
The answer lies in understanding the fundamental nature of innovation. Within the realm of industrial development and economic growth, most innovations broadly focus on increasing efficiencies. This is because the technocratic capitalist society we live in both praises and promotes efficiency as a norm. This highlights the dominance of “instrumental rationality” (which the sociologist Max Weber described as a means-end rationality). It pursues efficiency as a goal whilst neglecting an evaluation of its outcomes. So, the concept of efficiency is imbued with a deferential quality; it is perceived as rational and unassailably objective, no matter the cost.
In technocratic societies such as in our digital era, it is technical efficiency that has become the ideological force shaping the economy and society. As a result, despite widespread anxiety about technological unemployment, the labour-saving technologies of the current era are not just accepted as normal but also as rational and desirable.
An excessive focus on automation technologies
What this leaves us with is an excessive focus on automation technologies, particularly in relation to the dominant technology of our era—AI. The fact that AI is software also means that a fair amount of change is taking place in the background, both literally and figuratively behind the screen—not just computations out of sight, but also an AI black box. Several analysts and scholars argue that its coded internal logic secretly cultivates technological unemployment as a result of market forces. For example, Silicon Valley analyst and technology writer Tim O’Reilly cautions that “the economy is running on the wrong algorithm”, with the pervasive “master objective” to expand corporate profits at all costs, even at the expense of jobs.
At the same time, the economic historian Carl Frey has linked the growth of monopolies in the digital economy to an increase in automation technologies. He cites the profit-incentivised expansion plans of companies as the source of labour-saving innovations. He connects the pursuit of monopoly status to the increased adoption of automation technologies. This follows from the fact that wider profit margins increase prospects for market domination.
Relatedly, two different teams of MIT economists highlight jobless economic growth as a significant problem that has emerged as a result of 21st-century automation technologies. Eric Brynjolfsson and Andrew McAfee note a “great decoupling” between economic growth and job creation. Meanwhile, Daron Acemoglu and Pascual Restrepo argue that current innovations result in insufficient productivity gains. Due to a preoccupation with inventing technologies that prioritise automation, this has a “displacement effect” on jobs. Acemoglu and Restrepo specifically attribute the emergence of this “wrong kind of AI” to market failure.
The impact of technological unemployment
When the AI winter ended just over a decade ago, the digitalisation of economies became a trending issue. Suddenly, the middle class came into view as AI started encroaching on what were once considered stable jobs that supported decent livelihoods. Not only did this lead to a wave of automation anxiety amongst the general public, but it also spurred a number of studies with a specific focus on labour market dynamics. These studies predicted the nature and scope of job losses, including the kinds of jobs that would be lost to automation and the people that would be affected.
At the time, the widely cited Frey and Osborne study on the impact of computerisation in the US labour market predicted that nearly half of American jobs (47%) were at risk of automation. Based on the statistical model deployed by the US study, a local study by Stellenbosch University’s Daniel B. le Roux argued that the jobs performed by 35% of South Africans were susceptible to automation. It argued further that the jobs most at risk were “low and medium-skilled white-collar occupations” in the private sector and that there was a prevalence of “previously disadvantaged” South Africans in these occupations.
As part of a PhD study that examined the intersection of innovation and inequality, I went in search of the evidence to corroborate the claims made by these studies. This led me to the banking sector, which took the lead five years ago as the most rapidly digitalising sector in the South African economy. The surge of digitalisation that swept through the banks in 2019 resulted in the automation of internal operating procedures, as well as the migration of customer interaction to mobile and online platforms. While this led to efficiencies for both the banks and their customers, a significant after-effect was the prospect of technological unemployment for large numbers of bank staff.
Concerns about job losses in the banking sector came to a head in September 2019 when SASBO, the finance trade union, threatened a national strike, which was ultimately halted by a court interdict. At the time, the media reported that the digital re-alignment strategies of South Africa’s top four banks resulted in a number of clerical support and client services jobs becoming obsolete. This resulted in the loss of approximately 6,000 jobs. Meanwhile, in the ensuing years, banks have been quietly reducing their headcount.
What was curious about the interviews that I conducted with banking sector representatives is that they spoke about technological unemployment in abstract terms. They referred to ‘functions’ that were becoming automated. They refused to respond to questions about the demographics of those whose jobs were most at risk and/or to confirm the figures quoted by the media. Industry reports, on the other hand, revealed that large numbers of Black women, who were corralled into clerical support jobs, were the majority in banks. At the same time, nearly two-thirds of bank staff lacked post-matric qualifications. This highlighted a correlation between Black women with low educational attainment in low-skilled jobs and those who were at the battlefront of job losses during this peak digitalisation period.
Intriguingly, the higher up the ladder I went, the vaguer the responses were to questions about the demographics of technological unemployment. In fact, the leading executives counteracted my questions with oblique references to their banks’ transformation and BEE targets.
Despite these claims, the distribution of labour in these institutions reveals that banks have been slow to embrace transformation in any meaningful manner. Employment equity targets have historically been addressed by hiring Black women in proletarianised white-collar jobs. These are now becoming redundant. Meanwhile, their reluctance to respond to questions about the demographics of technological unemployment exposed the banks as employers embedded in the logic of technological rationality. They have absolved themselves of any responsibility to deal with the human cost of automation.
The banking sector can be used as a pacesetter for the white-collar economy, especially in relation to its significance for the expansion of the Black middle class. This makes it clear that technological unemployment represents a new threat to South Africa’s transformation. Job losses as a consequence of the adoption of automation technologies have definitely taken place. However, sensitivities around who is affected create an intentional silence about higher attrition rates amongst certain demographic groups. These challenge the dominant narrative on transformation.
The platform economy to the rescue?
Based on the idea that the AI revolution not only eliminates jobs but also produces new ones, there is incessant promotion of the platform economy as a new frontier for job creation. This is in line with an ongoing emphasis on skills development. In this respect, there are some outlandish claims made about the platform economy. These include the argument that it has the potential to lift South Africans out of poverty and that it “is the pathway that will drive growth and jobs across Africa”.
Proponents of this view promote the platform economy as a vehicle to develop South Africans as a nation of micro-entrepreneurs. While this may invoke romantic notions of flexible working hours and self-directed work that is personally satisfying, several studies challenge this perception. They expose the emergence of a precariat in the platform economy. In fact, the exploitation of workers in the platform economy is so extreme that it’s given rise to an explosion of precarity scholarship. This characterises gig economy jobs as work that is evidently only accepted under economic duress.
In general, platform work is grouped into location-specific gig work (like app-driven deliveries) and online crowdsourcing work (outsourcing tasks or projects to people via the internet). My study of the tech start-up sector reveals that the most common labour platforms to have emerged in South Africa’s digital economy are those that offer location-specific gig work. This also happens to be the most exploitable form of work in the platform economy. As might be expected, “the uberisation of work” has entered labour market discourse as a pejorative term to describe the race to the bottom.
To put it crudely, what is being presented as the solution to South Africa’s unemployment crisis is the digital economy’s version of kitchen girl and garden boy jobs. This is an insult to South Africans who have struggled so hard for freedom from oppression.
Online crowdsourcing work, which is not location-specific, tends to require higher skill levels. However, notwithstanding the skills requirements, there is growing evidence of an emerging hierarchy in the global labour market, indicating greater exploitation of workers in the Global South. Evidence has surfaced of non-Western workers being channelled into less skilled and poorly remunerated crowd work referred to as microtasks. For example, a Time Magazine investigation found that OpenAI paid Kenyan crowd workers less than US$2 per hour to label problematic content that was used to train ChatGPT to avoid offensive language.
We need better solutions for a better future
It’s absurd to continuously preach educational attainment and skills development for a better future in digitalising economies that predominantly produce precarious work. Prioritising education and skills within populations are important goals for societies to strive towards. However, this should be done for self-actualisation. It is ironic to link it to employment, given that the future will undoubtedly be one where there are fewer jobs available, as a structural outcome.
One solution that delinks livelihoods from employment is the introduction of a universal basic income (UBI). It’s a remedy that is popular amongst tech entrepreneurs, who understand that technological unemployment is becoming a permanent feature of society. Everyone from Elon Musk to Sam Altman supports the idea of a UBI.
Moreover, as the efficiencies created by technology create the opportunity for better work-life balance, another idea that has re-emerged is that of the four-day working week without a reduction in pay. Similarly, we might consider the revaluation of pay for low-paying jobs that are difficult to automate, such as in the pink economy (care work), which happens to be where some of the most vulnerable workers in our society are located.
Ultimately, we need to embrace the idea that technological unemployment is here to stay, as well as the fact that we can do better by those left behind. South Africans are already living in a society where half the working-age population is unemployed. Our response has been to endure the discomfort of living in the most unequal society in the world. It’s time to move forward differently. Not only is it time to experiment with new ideas, but it is also time to move away from the prejudices of the past as we reflect on what kind of future society we want to build.
Fazila Farouk is a digital economy specialist based in Cape Town. She writes about the intersection of innovation and inequality.
0 Comments