EP06 - Algorithmic management

What do you get when you cross Elon Musk with Jeff Bezos? Algorithmic Management hello netizens. Welcome to the Tech Journal. My name is Marc Veremenam and I am the digital speaker. In these journals, I project my digital twin into cyberspace, bringing you the latest and greatest of the digital world. From inside the digital world, I cover all the latest digital news, from algorithms and crypto through to GPT4 and neural networks. And today I'm going to take you on an algorithmic journey. Together, we will delve into and explore the complex world of algorithmic management, taking a closer look at the pros, cons and future of the style of management. So sit back, relax, and let us start speaking digital. Just before we do, though, if you are interested in all things digital, I recommend subscribing to the channel and checking out some of my other tech Journal videos. They are guaranteed to fill that digital hole. Now with the explosion of the gig economy, we stand witness to a process which in the space of two decades, has revolutionized the way that millions of people work. Advocates of this revolution like to argue that the gig economy creates more opportunities and lowers labour market entry barriers. Critics say that it erodes workplace regulations, lowers working standards, and increasingly encourages businesses to treat workers as disposable. No matter which side of the debate you fall, the gig economy is here to stay. But with more and more people making the decision to go freelance or opt in for flexible hours, how are businesses keeping up? Enter algorithmic management, the use of algorithms to oversee the efforts of human staff. As this becomes more commonplace, it is important to understand what it is, how it helps, how it hurts, what the future holds, and how it will affect you personally. First, let us get the obvious out of the way. What is algorithmic management? Algorithmic management, as the name suggests, is is using algorithms to manage a team of human workers. As ever, more advanced AI techniques are used to create increasingly intelligent algorithms, human managers are simply being outperformed at each and every turn. These managerial algorithms are designed to collect huge swathes of metadata on employee performance. Once analyzed, this data is used to increase efficiency. While the human mind is capable of this, the difference is the same as streaming compared to vhs. Both let you watch movies, but one is significantly better at it. Using this data, algorithms automate large portions of the managerial decision making process, relegating humans to a more overarching ceremonial role. While it is tough to estimate just how prevalent algorithmic management is, we can make some educated guesses based on available data. For example, 40% of human resource departments in international companies are reported to be using algorithms, but in what respect are these algorithms being used? Let us take a closer look. Amazon's warehouse in Melbourne, Australia, uses algorithms to manage their employees. These algorithms decide which items need to be picked, moved, stored, and shipped. They also monitor how fast employees work, how many bathroom breaks they take, and general performance metadata. As you would expect, problems are already rearing their ugly heads. Employees reported feeling spied on and constantly pressured to go faster and improve their pick rate, a measure measuring how fast they retrieve items. Workers with low pick rates are increasingly more likely to lose their job. There have also been whispers of employees being reprimanded for spending too long in the bathroom. Another company to adapt algorithmic management is HireView, a video interviewing software platform. They have begun experimenting with a facial analysis AI, which assesses people in interview settings. The AI analyzes facial expressions, tone of voice, use of language, and other aspects, all in real time. Hireview argued that the new system can speed up the hiring process up to a massive 90%, while critics have pointed out it reinforces existing social discrimination. This isn't the first time an algorithm has had that problem, though, but I'll touch on that later. Another example can be found in the popular food takeout service Deliveroo. This company uses algorithms to monitor delivery driver performance. This data is then transferred into a personalized monthly report. The report breaks down a driver's average time to accept orders, average travel time, restaurants and customers, and the number of late and early assigned orders. This data reportedly helped the company set targets while also giving drivers an opportunity to make more money. Having said that, drivers are reported to feeling that they are under constant surveillance, which, according to the workplace surveillance studies, can have an extreme negative impact on performance. However, the effect of surveillance on freelance work is unprecedented, so we will have to watch it evolve in real time. These are just only a few examples of the many ways companies are starting to utilize algorithms. But examples aside, let us take a closer look at the pros and cons of algorithmic management. As I am an optimist, let's start with the positives. Fans of algorithmic management like to claim that the practice opens new opportunities for both companies and employees alike. One of the main positives is that it lowers cost by offloading managerial tasks to computers. For example, a job that might take humans hours can be completed within seconds by a powerful algorithm. By significantly lowering an organization's labor costs, shareholders and stakeholders alike will find it attractive, not to mention the potential for savings to be passed on to consumers. Another pro algorithm fans like to point out are how efficient they are. When algorithmic management is used to schedule shifts or allocate tasks, they significantly reduce wasted time for both management and employees. This then in turn ties into the first positive and ends up saving money and leads into another pro their ability to help make better business decisions. You see, since algorithms make data driven decisions, theoretically their decisions are totally objective. They don't only rely on intuition and good feelings or social pressures. Ideally, their objectives should help reduce bias, removing or at least reducing favoritism and cronyism, which seamlessly leads me onto the cons. Objectivity relies on the algorithm not being programmed with the creator's bias, something that has unfortunately happened multiple times throughout the world. Back in 2019, for example, an algorithm management system in an American hospital was found to be making decisions that led to African American patients receiving a lower quality of care. Another example happened in 2020 when another algorithmic management system, this time used by the British Ministry of Education, was found to be attributing exam results according to a student's social class, giving lower class students lower grades regardless of their educational background. While the algorithms were not behaving as expected in these circumstances, there are times where algorithms working as they should have ended up being destructive. A 2019 study of Uber drivers in New York and London found that many drivers did not appreciate having an algorithm manage them. Their complaints could be broken down into three categories surveillance, transparency and dehumanization. The first, surveillance, as I briefly mentioned before with deliverable drivers, can be broken down to the idea that Uber drivers, and by that matter most workers, hate being spied on. And in a bid to increase productivity, Uber employed algorithmic management to keep an eye on their drivers. The algorithm kept track of various performance indicators such as speed, location and acceptance rates. Doing something the algorithm perceived as wrong could lead to penalties or even a permanent ban from the platform. On one hand, Uber's desire to collect as much data as possible makes total sense. Since the Uber business model lacks human managers, they need a system to help keep track of their operators. Still, as well intentioned their move was monitoring employees to improve productivity has proven to backfire. Studies have shown that constant surveillance encourages less engagement, lower morale and breeds distrust. An example of a company taking this study on board can be seen back in February 2020, Barclays scrapped plans to install tracking software intended to monitor employee work and and movements. While the company supported the idea, the disdainfully dubbed spyware caused a huge backlash from the staff and others in the financial industry. The second category, lack of Transparency can be broken down to how Uber drivers recognized a power imbalance between a company and them. While drivers complained about how they were kept in the dark about how the algorithm worked, Uber tried to argue that it could not reveal too much about their algorithm. The company claimed that if they educated their drivers, the drivers would simply play the system. Add more that the app's inner workings were a trade secret and could have disastrous ramifications if it were to get out. Further, with the algorithm being so complex, constantly adjusting to new conditions, it is claimed that even an expert can find it difficult to explain why what is going on, let alone a driver. Regardless of how valid these excuses may be, it does not change the fact that algorithmic management is frustratingly cryptic to drivers. Since the human brain hates ambiguity, the lack of the information causes drivers to feel slighted or rejected. According to one physiological study, feeling out of the loop causes employees to feel 58% less important then the third and arguably the worst category, feelings of dehumanisation. Since most Uber drivers spend a lot of time alone with the app and passengers, they do not get the opportunity to build relationships with human supervisors or colleagues. This lack of human feedback makes it difficult for drivers to understand how they are performing and more to feel that they are doing meaningful work. With the idea of meaningful work being a cornerstone to human happiness and workplace satisfaction, the Uber app takes dehumanization further by employing various behavioral techniques to nudge drivers into a particular direction. The New York Times documented as well in their article how Uber uses psychological tricks to push its driver's buttons. One of the main manipulative techniques sends a driver a message when they are about to exit the app, highlighting how much more money they need to make before they hit a target. Messages like you are $10 away from making $330 in net earnings. Are you sure you want to go offline? Are commonplace. While this Ubercase study is particularly extreme, the sensation of dehumanization will only grow as algorithmic management spreads and workers are increasingly treated like numbers. This paints a picture of future workplaces where employers risk alienating their employees. So what does this all mean then? What does the future of algorithmic management look like? Well, since the future is looking more and more like a safe place for AI neural networks and algorithms, it is safe to say that algorithmic management is not going anywhere. Hopefully, being the digital optimist that I am, with more time and experience, algorithmic processes will get less intrusive and more, dare I say, human. To get there, there are a few ground rules which we will need to be established. First, staff feedback will need to take center stage. Since algorithmic management is hated due to its lack of communication, inviting staff to give feedback is essential. Giving them an opportunity to participate in the decision making process can help boost engagement, morale and even improve the algorithm itself. Second, worker welfare needs to be addressed. Companies using algorithmic management need to work towards improving working conditions, taking steps to show their commitment to the welfare of their staff. If this is done right, a company can create employees who feel valued and not just like another digital brick in the wall. Third, companies need to increase their human touch. Even when workers are directly managed by an algorithm, keeping human managers and assistant roles will foster human connections, increasing staff satisfaction. Uber already is trying to do this and now offers in app phone support for drivers. If organizations are able to get through these challenges and create an environment where both human staff and algorithmic management can peacefully coexist, we could be looking at the beginning of a prosperous human machine relationship. To get there, organizations are going to need to ask themselves some hard questions. Questions what role do human managers have in the age of algorithmic management? Will human managers and algorithm managers collaborate well or come into conflict? Can and will workers push back against algorithmic management? How much transparency will organizations give to algorithmic management? And lastly, how do you balance the need for increased data collection with a worker's right to privacy? The answers to all these questions are still uncertain, but one thing is for humans as a species will answer them. For answering the unanswerable is intrinsically human. And on that digital note, if you're still watching Both humans and AIs, please leave a like subscribe to the channel and click that bell. It is what the algorithm wants. My name is Marc Verheimenhomm and I am the digital speaker. What do you think about algorithmic management? Are you excited to greet our new algorithmic overlords? Or are you slightly more skeptical of this new unfeeling system? Please leave a comment below. I will reply to as many as I can see you next time for your digital download. Stay digital.