The robots are here. What now? Stephen J. Frenkel and Tom Barnes explore the possibilities for workers and employers.
The debate about artificial intelligence and automation is often presented as a contest between optimists and pessimists. One side predicts mass unemployment as machines replace human labour. The other points to earlier technological revolutions and argues that new jobs will eventually replace those that disappear.
Both approaches risk missing the more important question. Technology does not determine the future of work by itself. What matters is who controls its introduction, how work is reorganised around it and what power workers have to influence the process.
Our research into robotics in an Australian supermarket warehouse illustrates why. Between 2018 and 2022, Woolworths, Australia’s largest supermarket company, introduced a highly automated distribution centre in Melbourne. At its heart were dozens of large robotic arms known as Robotised Palletisers. Their job was to take boxes of groceries and stack them efficiently onto pallets destined for individual supermarkets.
What happened next was considerably more complicated than the familiar story of machines replacing workers.
It looked like a textbook example of automation. The robots could calculate how boxes of different dimensions should be assembled on a pallet, rather like playing a three-dimensional game of Tetris. They were faster and more precise than people performing the same task. They also removed some of the physically demanding lifting associated with warehouse employment.
But what happened next was considerably more complicated than the familiar story of machines replacing workers.
Automation is a choice
Technology arrives in workplaces through decisions made by people with particular objectives.
For Woolworths, automation formed part of a strategy to increase warehouse capacity, reduce distribution costs and improve its competitive position. The new warehouse could handle almost twice as many product lines as the facility it replaced.
Management initially envisaged operating it with around 120 employees, compared with approximately 700 at the old warehouse.
Most workers at the previous facility lost their jobs. Entry-level wages at the new site were also 12.5% lower.
The technology therefore undoubtedly increased management’s power over labour. But this was not simply the consequence of some irresistible technological logic. Management had chosen a strategy that concentrated heavily on replacing labour rather than developing workers’ capabilities to work with the new technology. Other choices were possible.
We should ask what businesses are choosing to do with it, who benefits from those choices and who gets a voice in making them.
This distinction matters far beyond one warehouse. As AI and robotics spread through economies, debate should not simply ask what technology can do. We should ask what businesses are choosing to do with it, who benefits from those choices and who gets a voice in making them.
The robots had power too
There was another complication. Management could buy the robots, but that did not mean managers could completely control them.
Woolworths relied on Vanderlande, the specialist technology company supplying the automated system, whose engineers installed, monitored, maintained and repaired the machinery. The technology created a new group of skilled workers whose knowledge became essential to operating the warehouse.
The robots also dictated the rhythm of work.
Once integrated into the warehouse system, they determined the speed and sequence of activities around them. Human workers had to keep pace with the automated system.
Yet managers themselves could not simply order the robots to work faster.
As one manager put it, the system was “governed by what the system can do”.
This is an important and frequently overlooked aspect of automation. Technologies are created with particular capabilities, rules and limitations built into them. Once installed, those characteristics constrain managers as well as workers.
Automation can therefore increase management control over labour while simultaneously reducing management’s control over parts of the production process.
Humans returned
The limitations became particularly obvious when demand increased.
Robots sometimes dropped products or packed them incorrectly. Fragile or awkwardly shaped goods could cause problems. When a fault occurred, the system had to stop and humans intervened.
A manager could not tell a robot to operate at 200 per cent of its normal speed. But management could bring another 20 workers onto a shift.
More fundamentally, automated systems had limited capacity to respond quickly to sudden increases in demand.
During the COVID-19 lockdowns, supermarket demand surged. A warehouse designed around an expectation of around 120 workers ended up employing nearly 450. Why? People were more flexible than machines.
A manager could not tell a robot to operate at 200 per cent of its normal speed. But management could bring another 20 workers onto a shift.
By late 2021, after exceptional pandemic demand had subsided, the warehouse still employed roughly 300–350 people: far fewer than the old warehouse, but dramatically more than originally planned.
Automation had displaced jobs, but it had not eliminated human labour.
Workers wanted access to the technology
Perhaps the most surprising development concerned workers’ attitudes towards the robots.
Many recognised that automation threatened employment. But they also welcomed the reduction in strenuous manual handling and saw robotic technology as interesting and potentially valuable.
Their complaint increasingly became not that robots existed, but that workers were excluded from working with them.
Initially, specialist engineers had almost exclusive access to the robotic areas. Manual workers were classified as lacking the qualifications necessary to work there.
Workers do not necessarily resist technology. They may instead fight over who gets access to it, who develops the skills associated with it and who captures the benefits it creates.
Workers began demanding training and job rotation. They wanted opportunities to monitor robots, solve problems and acquire new skills rather than spending every shift undertaking repetitive manual picking.
Trade union representatives took up these demands.
Eventually management conceded greater access, allowing more warehouse workers to rotate through automated areas and learn from specialist technicians.
This reveals something important about worker power in technologically changing workplaces. Workers do not necessarily resist technology. They may instead fight over who gets access to it, who develops the skills associated with it and who captures the benefits it creates.
Technology does not abolish politics
Our case suggests that the future of work cannot sensibly be understood by counting the tasks machines might theoretically perform.
Workplaces are social systems. Technologies enter organisations where managers, workers, unions, specialists, suppliers and machines themselves interact.
Power operates throughout this process.
Which path we follow will not be decided by robots or artificial intelligence.It will be decided through choices about ownership, management, technology design and regulation.
Management has the power to decide which technologies to purchase, which jobs disappear and how work is organised. Technology suppliers gain influence because firms depend upon their expertise. Machines constrain what both managers and workers can do. Workers retain their own capacity to negotiate, cooperate, resist and demand access to new opportunities.
The result is rarely the frictionless automated workplace imagined in corporate presentations. Technological change is experimental, contested and permanently unfinished. That has wider implications for today’s AI debate.
If businesses approach technological change primarily as an opportunity to eliminate labour and reduce costs, automation is likely to increase inequality and weaken workers’ bargaining position.
But technology can instead be used to augment human capabilities, eliminate dangerous or exhausting tasks and provide workers with new knowledge and opportunities. Which path we follow will not be decided by robots or artificial intelligence.It will be decided through choices about ownership, management, technology design and regulation.
As Frenkel argues in a related paper, these decisions will require informed workers organised to effectively negotiate productivity gains and retention of meaningful work.
The real question about the future of technology and work is therefore not whether the machines are coming. They are already here. The question is who gets to decide what we do with them.
This has been produced with assistance from ChatGPT from an original academic article here.
