Senseless work. Can artificial intelligence support it?

Why technology doesn't take our jobs away?

People have long been accompanied by both hopes and fears that technology, instead of relieving them at work, will completely replace them and make them redundant. From a historical perspective, it is worth mentioning, among others, the emergence of the Luddites in England at the beginning of the 19th century, in the middle of the industrial revolution. The Luddites protested against the practice of using machines to produce lower quality goods for less pay for workers, which still, let's admit, is a quite contemporary argument.

In 1930, the well-known economist John Maynard Keynes predicted that in 100 years we would be so rich as a society that it would be enough to work no more than 15 hours a week to maintain a high standard of living. A few years later, in 1946, Austrian psychiatrist Viktor Frankl, who dealt with the issue of the meaning of life in his research, argued that it was the advancing automation that significantly shortened the working time. However, today we know that both were significantly wrong in their predictions related to the length of work, although they were right about automation.

So why, despite the intensive technological development aimed at making work easier and more efficient, do we observe the opposite phenomena? Why is declared job satisfaction low, there is a lack of a sense of purpose coming from it, and instead, ubiquitous stress, rush and so-called hustle culture dominate? What went wrong?

Or maybe work doesn't make sense after all, about deprofessionalization

The answer to these questions may be the research of the American anthropologist David Graeber who died in 2020. In his book Bullshit Jobs, he argued that many jobs are actually unnecessary and exist only to maintain the illusion of employment and feudal order. An example pointed out by Graeber may be middle managers managing teams that organize themselves perfectly and do not need supervision.

Graeber's research is groundbreaking and even controversial insofar as it touches on the social taboo that work has become in late capitalism. Currently, we talk a lot about how artificial intelligence (AI) will affect the future of many professions, but much less about the current situation of many workers and the role technology plays here. Graeber draws attention to a number of systemic problems resulting in the proliferation of meaningless jobs.

One of such problems is the decline in the value of work as such and the decline in the importance of qualifications or professional competencies. The latter is professionally called deprofessionalization. Broadly speaking, it is a phenomenon when it no longer matters whether the person doing a job actually knows it. In the context of the ubiquitous culture of meritocracy, this sounds unlikely, but in reality it can be even worse - high competencies sometimes even hinder the performance of some meaningless jobs.

This happens, for example, when a young ambitious person after journalism studies takes up employment in "clickbait content factories", disguised as online news portals. Such a job is a striking example of meaningless work, and everyone involved is at least partially aware of this. The main goal is to try to capture human attention with appropriately constructed headlines, and then monetize it in the form of data about clicks, likes and other metrics, for example, time spent reading. In fact, solid journalistic skills or a sense of mission can only be an obstacle in achieving this goal.

But can generative AI tools be harnessed for such a game of appearances? They can and what's worse - they already are. At the end of October 2023, an article from The Guardian about finding the body of a young woman in Australia appeared on the news aggregating platform Microsoft Start. It was "enriched" with an AI-generated survey on... the cause of the victim's death.

Can AI support "meaningless work"?

In the last decade, and especially in the last year, we have observed intensive development of AI tools. This is a synergistic effect made possible by the development of many factors: Big Data infrastructure and neural networks, enabling the collection and analysis of huge data sets, the proliferation of smartphones and other digital devices that generate this data, and the availability of increasingly faster internet. At the same time, Silicon Valley tech companies are imposing a narrative that if a task can be automated - because AI allows it - it should be done, because it is an economic necessity.

This is a thoroughly mistaken, dangerous and harmful argument. However, it underlies the thinking that the development of artificial intelligence will lead to mass unemployment or make someone more proficient in this matter deprive someone less proficient of work. In this context, there is also a lot of talk about the need to improve one's qualifications - implicitly, to include the increasing role of AI in them. But is it really necessary?

There are already analyses suggesting that the costs of implementing artificial intelligence in organizations are high - so high that generative AI tools may face a cold shower in 2024. For example, experts from the Massachusetts Institute of Technology argue that implementing AI in companies still requires the cooperation of many departments, sometimes far-reaching reorganizations or changes in work culture - and this costs. It is not without reason that it has been known for decades in management sciences that technological development does not necessarily make us more productive - and certainly not everyone and not immediately.

Like the 19th-century Luddites, we can also ask whether the recipients of the product or service we produce are getting worse quality as a result of using AI. This has recently become the subject of a strike by screenwriters in the USA, who did not want to improve scripts after artificial intelligence.

Efficient inefficiency

In the same management sciences, there is also talk of so-called efficient inefficiency. This phenomenon occurs when technology is used to do things more efficiently that are fundamentally inefficient e.g. writing emails that lead to nothing. Generative AI models, especially large language models (LLM), can generate thousands of such messages in an instant.

If we look at Graeber's typology of pointless jobs, basically all the types of jobs listed there are suitable for being automated by AI. Will this make them make more sense? Absolutely not. Take, for example, programmers patching fundamentally flawed code. They are like firefighters extinguishing fires on the spot, who do not eliminate their causes - in a moment the fire will break out again, the code will crumble and so on in a circle. AI can do such work faster and probably cheaper. But does it make sense?

Interestingly, other pointless jobs from the so-called bullies group (involving harming or scamming others), like telemarketers or all kinds of salespeople - have already been partially automated in the form of persistent phone calls speaking in a robotic voice. Various reports on the impact of AI on the labor market agree that it is precisely such positions that have the greatest potential for automation.

Inefficient efficiency

What is invisible to the managerial lens and eye (and for technology) are human relationships and dependencies, which also permeate our work and have a significant impact on it. Anthropology knows many examples when entire families worked "inefficiently" in agriculture for years, resisting the use of various machines. The sense of this resistance was precisely the fact that the whole family is engaged in work together and each person can feel needed.

In modern times, inefficient efficiency is probably most visible in the example of remote work introduced en masse in lockdowns, when employees lost the opportunity for spontaneous meetings - so necessary for the flow of information in the company, not to mention mental well-being. Hence, digital creations such as the metaverse can for now become at most Zoom on steroids, because despite the development of these technologies, nothing can replace real contact with another person. Even if this contact takes a little more time and energy than generating a template email using ChatGPT or an avatar in VR.

The text was created without the use of generative AI tools.