Book Review of Darrell M. West’s The Future of Work: Robots, AI, and Automation

March 02, 2020

Review of Keynesian Economics, January 2020

See abstract

The conventional narrative about technology and how it will lead to ‘the future of work’ is best understood if we first assume that politics does not exist and that the economics profession not only has a good track record at predicting the future but is able to manage the economy (forget the bubbles of the last 20 years). In this vacuum, the narrative is simple: earth-shattering technological change is coming. Millions of ‘low-skilled’ workers are replaced by robots and unable to find employment or support themselves. Workers generally face a ‘skills mismatch.’ Society, driven by competitive companies that maximize profits, must adapt or face the consequences. The Future of Work: Robots, AI, and Automation by Darrell West serves as our wake-up call. 

The Future of Work considers these questions: How will technology change the economic, social, and political landscape? What will happen to workers, and their employer-sponsored benefits, when companies replace their jobs with artificial intelligence (AI) and robots? The book is divided into three parts. The first is an assessment of the growing prominence of technology. The second explores how the labor market and society need to adapt, and the last part evaluates the political climate in the United States. 

The book starts off by examining how technology is rapidly changing the world today. To make that case, the author looks at the spread of and different application for robots, AI, and internet connectivity into different facets of life. The author documents the rise in the number of industrial robots in use, the impressive abilities of fledgling AIs, and the promise of a superfast 5G-connected world, among other things. 

The picture that develops implies that these technological changes are widespread. But their effectiveness and the rapid pace of their adoption is never quite demonstrated. We are supposed to assume that a successful but limited test of a technology in a specific context means that the technology is ready for widespread adoption. For most of the examples cited, this is not convincing. Self-driving trucks perform well under closely monitored conditions, but that does not mean they could function well in real-world conditions, or that they could perform the variety of tasks truck-drivers are responsible for (for example, securing loads, fixing mechanical issues, driving in poor conditions). The economic data available also do not support the conclusion that technology is rapidly changing the economy. High productivity growth would indicate rapid technological change, yet productivity has been extraordinarily low for the past 15 years, suggesting that advances in technology are not meaningfully affecting output (Baker 2017). More broadly, unemployment is low, suggesting that people, not robots, are needed in the economy. 

Even if rapid technological change might not be happening now, the author would argue that it will at some unspecified point in the future. This could explain why technological advancements are not yet showing up in the economic data we have available. This relies on the author’s analysis of the potential of new technology, based on information released by the companies developing it, which frankly is not sufficient. When a company promotes new technology, the book assumes that it must be transformative. This view is largely shaped by press releases, news articles, reports from consultants, and anecdotes. Questions on whether companies would want to hype their adoption of new technology, or why the future of work suddenly gained national prominence after the worst recession in 80 years, are not asked. 

In effect, the author is assuming that politics does not exist: companies’ behavior is neutral, competitive, profit-maximizing, and collectively, should guide policymakers; workers have unreasonable demands in the face of a changing world and cannot possibly satisfy business needs. This deficiency is well demonstrated by the combination of two popular arguments in favor of robots:

  1. humans will be replaced by robots, leading to massive unemployment; and
  2. robots are necessary to solve labor shortages and demographic crises.

Taken together, these ideas are incoherent. 

Outside this vacuum, political economy is a more useful tool in understanding technological change. Literature shows that companies use the prospect of new technology to do things like drive down wages, put pressure on unions, satisfy investors and signal to competitors, reshape policy, or seek subsidies. Technology billed as ‘disruptive’ might be misleading or impractical (much like in the dot-com era), yet used for political reasons. This ‘automation grift’ is easy to see. Theranos, an automated blood-testing start-up that collapsed in 2018, infamously exaggerated its effectiveness; low-paid workers masquerade as high-level AI at start-ups; Uber, like much of the gig economy it has come to represent, rose to prominence via lobbying, not by the merits of its business model or technology. The author does not consider examples like these. 

The second part in The Future of Work outlines how the labor market will be affected by technology, and then argues that the ‘social contract’ will need to be remade to accommodate it. The author describes how large firms have fewer employees than in the past, and how working-age men are employed at lower rates. The conclusion seems to be that workers are withdrawing from the labor market and/or moving to non-traditional work, supporting the idea that there will be an employment crisis. The gig economy is presented as a solution to this, and as a model for the future. However, these conclusions are far too simple. Too many other factors complicate the story on firm structure and labor market trends, like globalization, the decline of manufacturing, trade agreements like NAFTA, attacks on unions, and a myriad of social, legal, and political changes since the 1940s. The gig economy is also remarkably small – just 1 percent of total employment (Appelbaum and Rho 2018) – and conflated with outsourcing, contracting, and temporary work. 

Furthermore, this discussion about the future employment crisis does not adequately explain why profound inventions of the past – computers, electricity, plumbing, telephones – did not cause massive unemployment and actually created new jobs, nor is it explained why new technology will cause a crisis this time. More concerning, changes in the labor market both in the past and in the future are treated as if they are natural or predetermined, while society is presented as having no agency in either. It is clear that the book is talking about a particular future of work, which is elevated to the status of the future without explanation. This is a future where technology, controlled largely by businesses yet subsidized by the government, drives change and society reacts. 

In this vein, the book’s goal is not to create a new social contract in its preferred future – one that might include the working class, black people, indigenous peoples, and many others left out of the current social contract – but to tweak the existing one. If the book wanted to reimagine society it might reconsider the questions it is trying to answer: How should technology change the economic, social, and political landscape? How should work be transformed? 

Unsurprisingly, its recommendations to remake the social contract mostly miss the mark due to this deficiency. Since benefits are largely tied to employment, the book recommends a system of portable benefits – a 401(k)-type system for accumulating sick leave, for example. Parenting, volunteering, and other activities outside of ‘work’ would be commodified so that they could count toward these benefits. This system is contingent on the adoption of an ‘independent worker’ category of employment, which would formalize gig work and eliminate employment protections, like the right to a minimum wage. This marries the worst aspects of contract and traditional employment in exchange for a complicated system that would probably not provide adequate benefits, especially if run by employers. The risk is also that employers might gamify benefit accrual or add onerous rules. 

The other major policy suggestion in the book is to embrace the concept of ‘lifetime learning’ so workers remain relevant in a period of constant change. This would be more accurately described as ‘lifetime training.’ The main issue is that there is an assumption that technical change will mainly affect those with ‘low skills’ via skills-biased technological change. This is not proven (Schmitt et al. 2013). Intuitively, there are ‘high-skill’ jobs, for example in radiology, that could be automated more easily than, as another example, a custodian’s. And at least today, since employment of those with less than a high-school education has outpaced that of the college-educated, the data do not fit the conclusions. 

Instead of using education to create citizens able to adapt to change strategically, education becomes training that chases what businesses want over time. This is buoyed by anecdotes about a ‘skills gap’ where employers cannot find workers with the right skills – an idea that is supported by questionable evidence. More simply, employers might be seeking to pay workers less, which could be why tech companies promote STEM education. The need for ‘lifetime learning’ of this nature seems unconvincing at best. 

Other policy recommendations are more generous to workers – working less and more leisure, paid family and medical leave (albeit targeted), expanding the Earned Income Tax Credit, upping support for trade-related job loss (but notably not changing trade policy), a basic income (that might be contingent on eliminating all other benefit programs), and creating better individual retirement programs (which might channel public money to high-fee Wall Street accounts) – but the devil is in the details. 

The third part and conclusion of The Future of Work is somewhat detached from the problems it has been trying to warn us of. Apart from the issues du jour like ‘fake news’ and populism, recommended political reforms are generally consistent with liberalism: money in politics is bad, gerrymandering should end, everyone should be required to vote, the rich should pay higher taxes, and the electoral college should be abolished. The unique prescriptions, portable benefits and ‘lifetime learning,’ are reiterated and emphasized. Both would fundamentally shift power to businesses, which would not have to adhere to labor law or provide substantial benefits for the former, and for the latter, shift the costs of training – traditionally borne by employers – to the government or to workers via savings accounts or other finance schemes. 

The author does not consider the two most simple solutions to job and benefit loss: 

  1. the government should divorce healthcare and other benefits from work and provide those benefits universally; and
  2. the government should provide jobs to anyone that needs one – a jobs guarantee – along with required education. These policies are more straightforward and vastly more effective.

The lack of a discussion around these policies is a major blind spot of the book, but not altogether surprising. In this sense, the threads of the book – the reliance on economic orthodoxy (especially about deficits and debt), the focus on ‘low-skilled’ workers, the unwillingness to acknowledge the political economy of technology – are woven together to constrain choices and deliver recommendations that are underwhelming at best.

Although lessons from the past about technology should temper alarmism about the future, it’s impossible to know what technological changes lie in store. What we do know is that our political problems will likely be the same and reflect tension over inequality and power. The Future of Work’s main failing is that, by leaning on mainstream economics, it radically de-emphasizes society’s ability to shape and manage the future. Society will be able to decide who owns robots and their underlying intellectual property, as well as who keeps the profits they generate; it can keep people from getting too rich or industries too bloated; and it can manage the economy, set industrial policy, decide what technology is important, and solve pressing problems. These are the pertinent questions about the future. Even a future where human work is no longer needed at all is an optimistic future with the right policies in place. 

With the disastrous effects of climate change just over the horizon – a discussion of which is notably absent from the book as well – political power must be built up around these ideas. Technology in and of itself, as well as adapting society to corporate visions of future, are no solutions at all. 


Appelbaum, E. and H.-J. Rho (2018), Just one percent of workers do gig work as main job, secondary job or additional income,CEPR, available at: just-one-percent-of-workers-do-gig-work-as-main-job-secondary-job-or-additional-income. 

Baker, D. (2017), Myths of job-killing robots obscure real causes of inequality,available at: 

Schmitt, J., H. Shierholz, and L. Mishel (2013), Dont blame the robots: assessing the job polarization explanation of growing wage inequality,Working Paper, Economic Policy Institute, Washington, DC. 

Support Cepr


If you value CEPR's work, support us by making a financial contribution.

Si valora el trabajo de CEPR, apóyenos haciendo una contribución financiera.

Donate Apóyanos

Keep up with our latest news