Algorithmic Decision-Making in Service Work: Job Autonomy Torn Back and Forth

6 min read

Who’s in control?

Service workers have a deep desire for job autonomy. Generally speaking, job autonomy refers the possibility of determining the courses of action in relation to an individual's own work processes. The extent to which they are granted this autonomy is another question, the answer to which depends on the specific work content in services, the corporate strategies for realising the potential of labour power, and the associated use of technology within work processes.

Widely used Algorithmic Decision-Making (ADM) systems, as rule-based or learning technical systems that have an algorithm at their core as a decision-maker show various levels of intervention in the work of service employees and may be utilised to fulfil different corporate strategies, mostly in a specific relationship between control and optimisation.

Workers themselves want to retain control over their work as much as possible. Using ADM systems does not fundamentally contradict this aspiration – yet they call into question the decision-making authority over work content and conditions.

In my doctoral thesis, I explored these potential areas of conflict by examining the influence of ADM systems on the job autonomy of service employees in Germany who perform a particularly large amount of intensive interaction work, i.e., who are in direct contact with customers, clients, or patients.

Solidified polarisations, most lose ground

Quantitative analyses of the extent of job autonomy reveal a clear divide within the service sector, which speaks volumes about existing inequalities: A comparison between 2012 and 2018 data shows that the already autonomy-strong sectors (e.g., information and communication, finance and insurance, real estate, professional, scientific and technological services, education) are among the autonomy winners.

By contrast, sectors with weak autonomy (e.g., trade, logistics, accommodation and food services, healthcare and social services) have continued to lose ground, which leads to the overarching theme of increasing polarization. Overall, gains in job autonomy are rarely high for service workers, but the losses are often serious and affect the majority of employees.

But how is the deployment of ADM systems reflected in this torn autonomy situation?

Potentials for job autonomy, but …

The core of my empirical work consists of interview-based company case studies in outpatient care and bank advisory services. A holistic software solution for digital work processes, including ML-based shift and tour planning, serves as an ADM example in outpatient care. An integrated client approach management system for clients, which essentially provides product recommendations, sets an ADM example in banking services.

The company case studies show that the intervention of these ADM systems in the work processes of care workers and bank advisors has predominantly positive effects on job autonomy. In both cases, improvements in the respective services as well as relieving tendencies for workers are recognisable. In the example of outpatient care, contributions to the quality of care, improvements in workload, and a facilitation of interaction work are achieved.

In contrast, the changes in job autonomy may have ambivalent effects on bank advisors: It is true that the quality and quantity of advisory services are improved, and day-to-day work is facilitated. However, the work of the advisors is also fraught with more contradictions such as a strict time-boundness of product recommendations.

However, service employees retain the power of judgement and decision-making in areas of work that are important to them. These findings must be evaluated in light of the corporate strategy behind the deployment of specific technologies. If the company's objectives lie within the scope of process optimisation to support workflows – beyond the expansion of control mechanisms for the utilisation of labour power – there is far less danger for the shaping and structuring of autonomous service work.

In any case, and in view of workers' desire for job autonomy, excessive technology-based control of work performance and output appears to be an outdated corporate objective. Ultimately, the increasing deployment of ADM systems calls for more co-determination at the workplace if employees intend to shape their implementation and utilisation.