Should the Supply Chain Profession Adopt Ethical Automation?
Automation is transforming supply chains, but the societal cost in terms of inequality may be high
The advance of automation in supply chains is nothing new, but the current outcry over labor shortages is expected to accelerate the rush to replace humans with algorithms and machines.
The arguments in favor of automation are so well-established that the trend almost seems like a force of nature. However, just as Mother Nature is not always benign, automation has its dark sides. One of them is societal inequality, according to Professor Daron Acemoğlu, an economist at the MIT Department of Economics, and co-author of the book The Narrow Corridor: States, Societies, and the Fate of Liberty (Penguin Press, 2019).
At the MIT Center for Transportation & Logistics’s 2021 Crossroads conference, Acemoğlu explained how automation is a root cause of wealth inequality. He argued that this outcome is not inevitable if companies and governments make better choices about how automation is deployed.
Grim inequality picture
Acemoğlu’s analysis of how different demographic groups have fared over recent decades paints a stark picture of economic disparity, especially in the United States.
Over the 40 years after World War II, the wages paid by US private businesses to workers grew rapidly. However, the trend has slowed and flattened over the last 30 years, and there has been a corresponding fall in labor’s share of national wealth.
The decline is strikingly uneven. Acemoğlu looked at the economic fortunes of 10 demographic groups in America and found that, from the 1950s to the 1970s, real wage levels gained in tandem at about 2 percent per year. “This is the basis of the American dream,” he said. After the 1980s the curves diverge. Real wages for groups with relatively low levels of education fell, even as productivity increased. The wage increases captured by individuals with first degrees were relatively modest compared to the country’s growth rate. People with higher and specialized degrees were rewarded with a significantly bigger slice of the national cake.
This divergence pattern is not just an American phenomenon. And it is notable that middle-class occupations such as clerical and back-office work have declined in a broad cross-section of countries. However, “the US is distinguished by being a leader in inequality,” said Acemoğlu. “What is unique to the United States is very sharp declines at the bottom of the wage distribution.”
He believes that these trends are rooted in tech-driven transformation and the changing nature of work.
The robots are already here
Certain types of jobs have borne the brunt of these negative impacts, particularly where companies can use specialized software and industrial robots to replace workers.
On a macro level, there was a rapid displacement of workers due to automation during the four decades after World War II, said Acemoğlu. However, this trend was almost perfectly counterbalanced by the creation of new wage-earning opportunities. Over the last 30 years, the automation train has gained speed — but without providing replacement jobs at a commensurate rate.
Acemoğlu noted that automation and its underlying technologies are not monolithic; impacts vary with the types of technology used and how they are applied. It is especially important to distinguish between the automation of tasks previously performed by humans, and technology that increases human productivity and gives manual workers different tasks.
Take, for example, the spread of industrial robots. Robotics has helped American enterprises — notably in the auto and metals industries — to compete with foreign companies and raise productivity levels. But the economic fallout for many local economies has been devastating and stoked inequality nationally.
It can be argued that the rise of artificial intelligence (AI) represents the next wave of automation, but Acemoğlu points out that AI is a very different animal. It is a versatile technology platform that can be applied to many business models and organizational structures.
There is optimism that this highly adaptable, multipurpose platform will create lots of new opportunities for workers. Acemoğlu cites two reasons for being somewhat skeptical of this assumption. First, in organizations that are lead adopters, the emphasis tends to be on using AI to replace routine work. Significantly, hiring rates have slowed in these lead, AI-driven enterprises. Second, a small group of highly influential big tech firms account for one dollar out of every three spent on AI, Acemoglu said. These firms set the tone for how AI is rolled out, and again, the emphasis is on using algorithms to perform routine tasks.
Inequality is not inevitable
Acemoğlu terms this tendency to unthinkingly reach for automation as a tool for cutting headcount and associated costs as “excessive automation.” Is there a better approach that is acceptable to industries including supply chain that see innovations such as AI as essential to their future competitiveness?
The answer to this question has never been more important. As Acemoğlu points out, we are “at a critical juncture” made even more immediate by pandemic-related changes in the workplace.
He firmly believes that excessive automation is not a preordained solution that companies must choose. Moreover, Acemoğlu does not accept that this path delivers compelling benefits such as increased productivity and goods that are cheaper and come in more varieties. “I think the evidence for this is scant,” he said. For example, we are experiencing historically low productivity growth rates at a time when tech-based innovation is flourishing. The number of patents has increased fourfold today compared to the 1980s.
Job automation is a perfectly valid strategy that has fueled industrial revolutions. However, “we can have ethical automation,” Acemoğlu said. A more thoughtful approach to automation is not driven primarily by cost-cutting where “automation is the only game in town.”
In the education industry, for example, AI is being used to automate aspects of teaching such as homework. But there is a huge, unmet demand for individualized tuition that could be supported by AI and create new job opportunities for teachers. An example closer to home is the introduction of autonomous trucks. Automating the operation of trucks could result in mass driver layoffs, or the technology could release many drivers to take responsibility for other supply chain tasks.
Companies can make such choices — but not without the support of government. “Corporate responsibility will not appear miraculously; you have to force it with regulatory institutions,” said Acemoğlu. The prosperity-sharing models of the 1950s and 1960s were underpinned by regulatory frameworks.
However, in his opinion, today’s government policies actually encourage excessive automation. An example is biases built into the tax code. “If you lay off workers and hire machines to perform exactly the same task with exactly the same productivity you make 20% more profit because the government is subsidizing it,” he said.
Learning from the leaders
Germany, South Korea, and Japan have become world-leading adopters of automation without suffering the wage inequalities that are a feature of many other economies, partly because of their governments’ approach to automation. Acemoğlu said Germany has installed three times as many robots per industrial worker as the US but also has government programs that support the redeployment of displaced workers. Education plays a key role. For instance, two-year vocational training programs give German workers new skills and make it easier for them to enter non-production occupations.
“It’s an ecosystem,” said Acemoğlu, that facilitates automation but not at the expense of prosperity-sharing opportunities for workers. If Acemoğlu is right, other countries can develop similar models. However, such a shift requires companies willing to look beyond automation purely as a means to eliminate labor, and governments with the political will to support a more expansive view of technology’s societal role.
