The Paternal Algorithm

From Paternalistic Leadership to Artificial Intelligence: A Brief Sociological and Psychological Reflection

Looking at models of leadership, a question occurred to me: could Large Language Models (LLMs), or algorithmic systems more broadly, be turning into a digital father figure for humanity? At first it may strike you, too, as a laughable or even overreaching question. But when we think about it through the relationship people build with technology, it does not seem so strange.

Technology has moved far beyond being a mere tool. It guides us, narrows our options, and decides what we see, what we read and which answer we reach first. In some areas of our lives, in other words, it works like a guide, even an invisible authority.

That is exactly why Zeynep Aycan’s model of paternalistic leadership opens a useful door for this discussion. Aycan’s model highlights several dimensions: creating a family atmosphere, caring for the individual, involvement in private life, expecting loyalty and maintaining authority. Going over them again, I suddenly found myself thinking of the Netflix and Google algorithms. And then, of course, of my phone.

By nature we dislike uncertainty. So besides attaching practical value to knowledge, we also invest it with a sense of psychological security. “Knowledge is power.” And to make sure we are never left uninformed, we could list dozens of sayings like “Sharing information is a courtesy.”

This is in fact a simple observation, but its consequences for society and the individual are enormous. Uncertainty breeds anxiety, and anxiety automatically triggers a search for control. Many studies of leadership and culture have addressed this and used it as a parameter. Hofstede, for instance, used “uncertainty avoidance” in his cultural model to describe national cultures. This is precisely where paternalistic leadership steps in and says: “Don’t worry, I’m thinking for you.” That sentence reassures us, because someone is taking on part of the responsibility of thinking on our behalf.

LLMs and algorithms do the same for us. When we drive, they choose the best route. They filter the most accurate answers. They decide which content suits us best. They lighten our psychological and mental load, reduce the burden and fatigue of decision-making and, more importantly, make us feel safe. There is nothing wrong with this, until we lose the ability to make decisions for ourselves.

Someone who is constantly steered gradually looks less at their own inner compass. In psychology this is called learned dependency. It need not be dependency in the technical sense. The system suggests, you accept. The system filters, you trust. And one day you realise that you are not really deciding what you want; you are merely choosing from whatever is offered. You may not even want to bother with options any more, and may wish to have them reduced to one.

Belonging is one of the strengths of paternalistic leadership. It makes people feel that “you are part of this family.” The sense of belonging is indeed a powerful motivation and one of our most basic needs. Today’s big platforms do exactly this. Followers, community members, network connections… When you leave a platform, you leave not only an app but also a circle, a community.

Loyalty, too, holds an important place in paternalistic leadership. In the digital world our “attachments” are measured through statistical analysis. Our personal preferences are analysed and efforts are made to intensify them. For some companies loyalty is now as much a data set as it is a feeling. In a sense, they are trying to manage our emotions with data sets.

As for authority, in paternalistic leadership relations with the boss may be close and warm, yet the hierarchy is always preserved, and everyone accepts this as natural. Algorithms have a hierarchy too, but it is not as visible as in social life. We do not know on what criteria content is ranked for us. Nor is it transparent which information is brought to the fore. For LLMs, concepts such as forbidden, good and bad are nothing more than “tokens”.

When transparency weakens, the problem is no longer a matter of too few or too many options; it becomes a matter of this space being arranged by criteria we cannot see. In a sense, our alternatives slowly become invisible. We think we are really making a choice, whereas the options have long since been filtered hierarchically for us. And we shape our thoughts of the moment according to the options the algorithm presents to us.

At this point we also need to ask: compassion or self-interest? In Aycan’s model, paternalism is treated in two directions: one that genuinely puts the good of the employee and subordinate first, and another that puts control and outcomes first. Technology companies constantly oscillate between these two axes. On one side, the effort towards safety, countering disinformation and content filtering. On the other, the paradox of keeping users on the platform, collecting data and maximising profit. As users we live a similar paradox. We feel both gratitude towards algorithms and suspicion. We trust them and yet want to keep our distance. This dilemma reflects the general state of mind of modern humans.

Weber listed three types of authority: charismatic, traditional and legal-rational. Perhaps the time has come to add a fourth: algorithmic authority. It is invisible, it produces no charisma, yet it builds legitimacy over users through data.

This piece is, of course, not a condemnation of technology. But let us not move on without asking: can we really tell guidance apart from steering? Can we see the line between being protected and being controlled?

If LLMs, as a paternalistic leader, are always by our side, when will we truly grow up, and when will we make our own decisions?