Algorithms are being deployed to automate management tasks across an increasingly wide range of industries and environments. For example, Amazon, Uber, and UPS use them to monitor the actions and performance of millions of drivers and warehouse workers, while 7-Eleven, IBM, and Uniqlo use them to track sales performance or assess employee skills for retail employees.
This shift to "algorithmic management" clearly provides companies with greater efficiency and profitability. But does it also have unintended consequences and alienating effects, especially when it comes to workplace dynamics?
Surprisingly, very few researchers have considered this question systematically, which means that we have little data to help us answer this question. To address this gap, in our study, we rigorously test whether the impact of algorithmic management extends beyond worker productivity. Recently, we examined the impact of algorithmic management on prosocial motivation, which is an important driver of creativity, productivity, social interaction, and overall well-being in the workplace. In the process, we identified a particularly interesting and important gap: algorithmically managed employees are less willing to help or support their colleagues than those managed by people.
Companies that use algorithmic management need to be mindful of this issue and be wary of other negative effects that algorithmic management can have on the psychological and social dynamics of employees. Fortunately, as we'll discuss in this article, our research shows that companies can mitigate these effects by creating work environments that actively encourage social interaction.
Our research began with a field survey of people working in the transportation, distribution, and logistics sectors, where algorithmic management is common. This was the first thing we found that algorithmically managed employees were less inclined to help or support their colleagues. This trend persists even when we consider organization-specific factors (such as size or average employee tenure), job (such as management satisfaction or overall satisfaction), and personal characteristics (such as gender and income).
Next, we conducted a field experiment in collaboration with a German truck rental company to directly test the behavioral results of algorithm management. At the beginning of this experiment, we paid about 1,000 gig workers from a **workforce platform to create a slogan for the van rental company's social ** marketing campaign. Workers were randomly divided into two groups, one guided and evaluated by an algorithm and the other by a human. Once the staff completed the task, we asked them to give advice to others on how to create an effective marketing slogan, and then we measured their willingness to do so.
What we found was noteworthy that algorithmically managed staff provided about 20% fewer advice to colleagues than those managed by individuals, and that the quality of their advice was also lower. (Interestingly, there was no significant difference in the quality of the actual slogans presented by the two groups, suggesting that algorithmic management does not necessarily affect worker task-based performance.) ）
When we conducted field surveys of workers in the transportation, distribution, and logistics industries, we found that regular social interactions between workers can hinder the negative effects of algorithmic management. This suggests that companies can actively mitigate adverse effects by fostering an environment where employees can connect with each other and communicate meaningfully. This may involve initiatives such as providing common rooms, implementing team rotations, and organizing social events or joint leisure activities.
In another study, we randomized participants to one of two conditions. In one case, participants read a work environment in which the managerial task described in the scenario is to evaluate the performance of workers. In another case, they read about the work environment, where the administrative tasks described in the scenario are scheduling and work planning. We also manipulate whether management tasks (evaluation and planning) are performed by human managers alone or by human managers using algorithms (which is how algorithmic management is often implemented).
Interestingly, we observed a decrease in prosocial motivation for algorithmic management only when the focus task was performance evaluation, which tells us that algorithmic management does not reduce prosocial motivation evenly across all management tasks. The negative effects are especially pronounced when algorithms monitor and evaluate employee performance. Companies need to be aware of this impact. If they decide to rely on algorithmic management in performance reviews and other HR-related tasks, they should make an effort to integrate human managers.
But even when human managers are involved, our research suggests that the use of algorithms in performance reviews still carries the risk of negatively impacting prosocial behavior. In the study just discussed and another study that directly tested the effects of human engagement, the negative effects on prosocial behavior persisted when human managers used algorithms to assess employee performance.
Anticipating this, companies and managers need to proactively inform and involve employees in decisions about the use of algorithmic management. When employees are recognized in this way and participate as stakeholders in the design and implementation of algorithmic management, they are more likely to maintain prosocial behavior and are less likely to feel objectified.
Companies such as Haier, one of the world's largest home appliance manufacturers, have effectively implemented automated performance evaluation systems that allow employees to establish their own performance benchmarks and exceed algorithmically determined minimum targets. In addition, companies need to ensure transparent and conscientious communication about how algorithms are used and who has the final say in the decision-making process. IBM, for example, incorporates algorithms into its compensation decisions, but also clearly communicates to employees that the recommendations these algorithms provide can be ignored by managers.
There's no denying that algorithmic management offers companies many new opportunities to improve the way they work. But we're just beginning to understand the impact of this practice on personal well-being, collaborative behavior, and team dynamics, so companies should be extra cautious when starting to use it. In particular, they should actively work to mitigate the possible negative effects of algorithmic management on prosocial behavior, which is critical to workplace success at both the individual and collective levels. A balance needs to be struck here, and companies need to work hard to find it.
Digital transformation of enterprises