What's wrong with employees when their bosses are replaced by 'algorithms'?

Some workers who are worried about their relationship with their boss may have thought, 'I wish my boss was an algorithm or AI instead of a human being.' In fact, in recent years, not only end workers but also managerial jobs have been automated, and Uber drivers and Uber Eats delivery workers basically work according to the instructions of the algorithm. Robert Donoghue and others, who are enrolled in the doctoral course of the Faculty of Social Policy Science at the University of Bath in the United Kingdom, explain the problems caused by such an ``algorithm boss''.

Horrible bosses: how algorithm managers are taking over the office
https://theconversation.com/horrible-bosses-how-algorithm-managers-are-taking-over-the-office-191307

There are many fictional works featuring disgusting bosses, such as the 1999 crime comedy movie ' Restructuring Man ' and the 2011 black comedy movie ' Monster Boss .' However, in recent years, software algorithms and AI are increasingly responsible for managerial tasks such as screening candidate candidates, assigning tasks, evaluating worker performance, and deciding which employees to fire.

In fact, it is reported that Amazon has installed AI surveillance cameras in some delivery vehicles to detect safety problems such as driver's non-stopping and distracted driving. However, it is said that there are many malfunctions in this system, and there are complaints that AI's erroneous judgments affect salary assessments.

Amazon has installed AI-equipped cameras on delivery vehicles and has begun monitoring drivers 24 hours a day - GIGAZINE

by Trevis Rothwell

Relying on algorithms to manage administrative tasks and automating tasks that would take a long time for humans can result in significant business cost savings for companies. For example, Uber, which operates a dispatch service, has tens of thousands of employees, but there are 5 million end workers including drivers and delivery workers, and it is difficult to manage them manually. Therefore, Uber has succeeded in managing a huge number of people by leaving the interaction with workers to the algorithm .

AI systems are also effective in optimizing businesses. Uber has introduced a price fluctuation system called `` dynamic pricing '' that monitors the usage status and the number of empty cars in each region in real time and derives the optimal usage fee from the balance between supply and demand. This is possible because algorithms can monitor passenger demand in real time.

However, it has been pointed out that algorithms have various problems, and among them, ``bias in algorithms'' is the most talked about by journalists and researchers. For example, it is known that Amazon's `` AI system for evaluating candidate resumes '' was abandoned because it had a bias to favor men and unfairly underestimate women.

It turns out that Amazon's top secret AI tool was discarded because it was 'female disdain' - GIGAZINE

Another problem is that it is unclear what kind of process the machine learning algorithm trained with a large amount of training data outputs the result. For example, if an algorithm's boss made the decision to 'fire employee A', it would be difficult to know if the decision contained bias. Donoghue points out that not knowing the breakdown of the dismissal process makes it difficult for employees to defend themselves against abuse of power by claiming that this dismissal is unfair.

Donoghue is also concerned that the introduction of algorithms into employment relationships will lose the 'compassion for subordinates' that human bosses should have. Of course, not all bosses treat their subordinates with compassion, but at least the algorithmic boss has zero chance of being compassionate. Algorithms that pursue efficiency are not tolerant of employees who are busy raising children or new employees who are new to work and are not used to work, and try to find employees who suffer from illness or disability and seek solutions. neither.

``Companies may find great benefits in management algorithms, but the need to make profits is no reason to tolerate employee suffering,'' Donoghue said, arguing that developers, researchers, and labor Unions, politicians argued that we need to consider the impact of algorithms on employees.