Algorithmic Management: What is It (And What’s Next)?
The growth of the “gig economy” in recent years has revolutionised the way that millions of people work. Proponents argue that the gig economy gives people more flexibility and opportunities and lowers barriers of entry to the labour market, while detractors say that it erodes workplace regulations and standards while encouraging businesses to treat workers as increasingly disposable.
No matter which side of the debate you fall on, it’s clear that the gig economy is here to stay. But with more and more people signing up for these flexible and freelance work arrangements, how can businesses manage them effectively?
Enter “algorithmic management”: the use of algorithms to oversee the efforts of human workers. As algorithmic management becomes more commonplace, it’s important to understand what this practice is, the pros and cons of using it, and what the future holds.
What is algorithmic management?
Algorithmic management, as the name suggests, is the use of computer algorithms and artificial intelligence techniques to manage a team of human employees. By collecting massive quantities of data, in particular data about employee performance, algorithmic management seeks to automate large portions of the managerial decision-making process.
While it’s tough to estimate just how prevalent algorithmic management is, there are a few indications. For example, 40 per cent of human resources departments in international companies are currently using AI applications. Below are just a few ways in which algorithmic management has begun to enter the mainstream:
- The video interviewing software platform HireVue is experimenting with a facial analysis AI that assesses factors such as a candidate’s facial expressions, tone of voice, and use of language. HireVue argues that the new system can speed up the hiring process by 90 per cent, while critics say that it could reinforce existing societal inequalities.
- Workers at Amazon’s warehouse in Melbourne, Australia are managed by algorithms that determine which items need to be picked, moved, stored, and shipped. Employees say they feel pressured to improve their “pick rate,” a metric that calculates how many items are retrieved from the shelves every hour.
- Drivers for the food delivery service Deliveroo receive monthly personalised reports about their performance, including their average time to accept orders, average travel time to restaurants, average travel time to customers, and the number of late and unassigned orders. For example, the platform expects drivers to accept new customer orders within an average of just 30 seconds.
What are the pros and cons of algorithmic management?
Advocates of algorithmic management and the algorithmic business contend that the practice opens new opportunities and efficiencies for companies and employees alike. The potential benefits of algorithmic management include:
- Lower costs: Offloading at least some of your managerial tasks to an algorithm can help significantly lower the organisation’s labour costs. For example, activities that could take a human hours or even years to do can be accomplished within seconds by a powerful computer.
- Greater efficiencies: Algorithmic management can help schedule employee shifts or allocate tasks more effectively, resulting in higher worker productivity and less wasted time for both managers and employees.
- Data-driven decisions: Using algorithmic management means that you have a (theoretically) objective way of making decisions, rather than taking action based on unscientific “gut feelings.”
- Less bias: As a corollary, algorithmic management may also help reduce or eliminate human bias and favouritism. (Of course, this is assuming that there isn’t bias itself baked into the algorithm’s design.)
Despite these benefits, however, algorithmic management has often been deployed in ways that are at least controversial, if not downright detrimental to an organisation’s goals. A 2019 study of Uber drivers in New York and London found that many of their complaints could be classified into three major categories that highlight the issues with algorithmic management.
Drivers are aware that they’re constantly under the watchful eye of the Uber mobile app, which tracks performance indicators such as their speed, GPS location, and the acceptance rate of new riders. Taking the “wrong” actions can lead to penalties or even a permanent ban from the platform.
Of course, Uber’s desire to collect as much data as possible makes sense, since they can’t have a human manager in the passenger seat for every ride. Still, trying to monitor employees to improve productivity can actually backfire in the form of less engagement, lower morale, and broken trust. Barclays, for example, recently scrapped plans to install tracking software that would monitor how long employees spent at their desk, and what proportions of time they spent on each task.
2. Lack of transparency
Many Uber drivers feel that there’s a power imbalance between the app itself, which is constantly monitoring their performance, and the workers themselves, who receive very little insight into how it operates.
Uber argues that it can’t reveal too much about the algorithm, since the app’s inner workings are a trade secret that could damage the company’s profitability. In addition, the algorithm is so complex and dynamic, rapidly adjusting to new conditions, that it can be difficult for even technical experts to explain what’s going on.
Regardless of how valid these excuses may be, it doesn’t change the fact that algorithmic management can be frustratingly opaque to the workers under its thumb. Because the human brain dislikes ambiguity, this lack of information may cause employees to feel slighted or rejected. According to one psychology study, feeling “out of the loop” causes employees’ perception of their group standing to drop by 58 per cent.
3. Feelings of dehumanisation
Even worse than the lack of transparency, perhaps, is the accompanying feeling of dehumanisation. Most gig workers don’t have coworkers or managers (human ones, at least) to socialise with—only themselves, the app, and passengers who frequently arrive and depart throughout the day.
Without building a relationship with a human supervisor, it can be difficult for drivers to understand how they’re performing, or to feel that they’re doing meaningful work. In addition, the Uber app uses various findings from behavioural economics to “nudge” drivers in a particular direction, a practice which is well documented in the New York Times article “How Uber Uses Psychological Tricks to Push Its Drivers’ Buttons.” For example, a driver who is about to exit the app might receive the message “You’re $10 away from making $330 in net earnings. Are you sure you want to go offline?”
While the case of Uber drivers is particularly extreme, the sensation of dehumanisation can befall anyone subject to algorithmic management. As technology becomes increasingly prominent in the workplace, employees risk feeling disconnected and alienated.
For example, automated shift scheduling software is used to implement “just-in-time” scheduling, in which employees are asked to come into work (or leave work) based on how much there is for them to do. However, by destabilising workers’ schedules, practices such as just-in-time scheduling have been shown to increase stress, income uncertainty, and the risk of work-family conflicts.
What is the future of algorithmic management?
Because of the potential shortcomings discussed above, organisations considering robots as management need to tread carefully. For the best chance at success, companies interested in using algorithmic management should take actions such as:
- Asking for feedback: Part of why algorithmic management can be so hated is that communication only travels in one direction: from the algorithm to the worker. Inviting workers to give their feedback and participate in the decision-making process can help boost engagement and even improve the algorithm itself.
- Improving working conditions: Companies using algorithmic management should take steps to show that they’re committed to the welfare of workers, not treating them like a cog in a machine. Postmates CEO Bastian Lehmann, for example, has argued for several reforms to working conditions in the U.S., including a permanent national sick leave fund for independent workers and a national portable standard that ties benefits to the worker and not the employer.
- Using the human touch: Even where workers are directly managed by an algorithm, adding in the human touch where possible can go a long way. Organisations should retain human managerial and assistance roles and help people foster connections with their co-workers. For example, Uber offers in-app and phone support for drivers who have a question or who need help.
So what’s next for algorithmic management and other applications of AI in the workplace? In our article “Revising the ‘science of the organisation’: theorising AI agency and actorhood,” my co-author Danielle Logue and I theorise about the future of AI capabilities such as algorithmic management.
We note that we have already seen AI agents that can change their behaviour, make decisions independently, and evolve without human interference (e.g. “Working for an algorithm,” Curchod et al.). In particular, the ability of AIs to create other AIs (e.g. Google’s so-called “child AI”) will require us to rethink organisational design, strategy, and governance theories that do not account for AI agents in terms of their differences from human actors.
Moving forward, organisations using algorithmic management need to ask questions such as:
- What is the role of human managers in an age of algorithmic management? How will these two sources of authority collaborate or come into conflict?
- How can workers push back against problems and issues that they encounter with algorithmic management?
- What level of transparency will organisations commit to with algorithmic management? How can organisations balance the need for increased data collection with workers’ right to privacy?