In 2017, Alibaba, the Chinese tech giant, introduced a revolutionary retailing model. The plan was to establish a network of “unmanned supermarkets,” where customers could order and pay for their products on a computer screen, robotic arms would collect and present the items, and the customers could ‘grab and go’.

2017 was also the year that I conducted my first fieldwork in the city of Hangzhou in China. As a researcher with a great interest in the digitalisation of work and employment, Hangzhou, also known as “the Silicon Valley of China”, was an ideal field site for my research .

At that time, it was widely believed that by reducing the cost of human labour, Alibaba’s new retailing model would be time-saving and price-competitive, and eventually change the landscape of retail service in China.  If the model were to succeed, commentators had no doubt that millions of Chinese workers in the retail industry would become redundant.

We watched with interest, wondering whether this new retailing model would change the landscape of retail service in China, and beyond? And if that were to  happen, would millions of workers in retail become redundant?

Did Alibaba herald a revolution in retail?

Three years later, I revisited Hangzhou, and the answers to some of these questions were becoming clearer. The outcomes and implications of Alibaba’s growth were more complex than many commentators had predicted. Alibaba-owned supermarkets had become one of the main players in the city’s grocery services. However, its supermarkets had shifted from unmanned to heavily manned: store assistants welcomed shoppers, gave out samples, tidied up shelves, and helped with selection and payment. Although all of these tasks could be performed by intelligent machines, Alibaba seemed to find human workers more suitable for their business operation, and not just because people may prefer to be served by people. As one of my informants told me,

The cost of setting up these machines is a huge investment… when a robotic arm is down, it takes two hours to send someone to investigate the issue; and then another few hours to send the parts for replacement; and then you have to fix it overnight before the business begins the next day. In comparison, when a worker is sick, you just call in another worker to take his shift. Problem solved.

Job replacement and Amazon in the UK

Five years on, I am exploring how these issues are playing out in the UK by analysing the UK’s answer to Alibaba: Amazon UK. I am learning whether the changes I observed in Alibaba are mirrored in Britain, and my research questions revolve around how Amazon workers interpret their job vulnerability as robots and intelligent machines become more common in the workplace.

Over the past few months, I have conducted interviews with frontline factory workers and their managers, and it is surprising to see the diversity of their opinions. Most of the frontline workers I spoke with believe that these digital technologies have actually created some jobs, and improved aspects of work in some instances. Some workers have moved from other warehouses, attracted by Amazon’s better pay, while others have highlighted that with assistance from robotics, their work had become less physically demanding.

However, Amazon’s managers have revealed another side of story: the implementation of robotics reduced the workforce by as much as 90 per cent  at individual job sites (at one warehouse,  60 workers were cut to six). Yet, as warehouses continue to be established, as Amazon’s business expands, a recruitment drive has camouflaged such redundancies. In 2021, Amazon UK created 25,000 new permanent jobs across corporate functions, R&D, and operations, and the number is expected to grow in 2022. Further, some Amazon workers are being upskilled to supervise the newcomers: both the machines and ‘human’ workers. Contrary to the observation made by my informant in China, the robots appear to be acting as cost-saving job producers, rather than expensive and unsuitable for retail contexts.

Reflecting on both case studies, my overall argument has emerged: technological innovations might automate some tasks in a particular labour process, but it does not necessarily mean the replacement of human workers. From the Chinese experience, the readily-supplied workforce gave human workers an advantage over robotics, which were more costly or time-consuming to be implemented. This was accentuated by the fact that retail workers were in such abundant supply.

In the UK, meanwhile, continuous business growth has outweighed the decreasing demand for human workers in each workplace’s labour process; as labour costs go up, intelligent machines could become more attractive, supplementing and even helping to grow the UK workforce. In both cases, job creation for ‘human’ workers is growing alongside the increased implementation of automation in the industry.

Technological unemployment is not inevitable

So what does this means for ordinary workers? I believe that it means we have good reasons to be hopeful.

My empirical findings suggest that the “technological unemployment” hypothesis (i.e., job loss directly caused by new technologies and innovations) is not inevitable. Some writers on the future of work make predictions based on whether tasks could be automated in a particular profession. I argue that this is, by itself, not a good indicator for the real risk of job loss. In order to understand the impact of automation on work and employment, it is absolutely essential to consider the political economy of technology in a particular workplace.

But staying hopeful is far from enough. Only actions will bring security.

For workers, giving voice to individual work experiences becomes more important in this new world of work, because it helps reveal the actual impact of technological transformation in the workplace. This might be achieved individually through social media platforms, or collectively via union representatives. And surely, researchers need to bring more information together, in support of evidence-based policy advocacy. Government policy is one of the key factors that will shape the future business environment and labour standards, which eventually determine how the implementation of technology is negotiated in the workplace whilst protecting the employment of a ‘human’ workforce.