Blessing or Curse? AI Dilemmas I.

Human development has perhaps seen two major milestones: the use of tools and the emergence of structured speech, which unlocked the possibility of communication.

But now, for the first time, we are encountering a technology that is simultaneously a tool and a communication partner.

This article attempts to provide a picture of our natural fears and their relationship to reality from a user’s perspective. It does not offer definitive answers, as those cannot currently be formulated with absolute certainty.

Until now, our tools only evolved incrementally: cars became faster, and industrial robotics made our factories more productive. At the same time, our machines did not possess even the slightest semblance of autonomy; as users, we could oversee every moment of their operation, and we never felt like we were losing control over them.

In sci-fi literature, numerous writings features automatons that gain consciousness. Naturally, the thrill does not stem from them being helpful, but rather from their desire to create their own society where humans are obsolete. At the same time, well-known works also speak of the possibility of cooperation. Isaac Asimov formulated the Laws of Robotics as early as the mid-20th century. But we can also remember HAL, the complex computer system from the brilliant book 2001: A Space Odyssey, where the shifting light of the red lens in the film version almost projected the rhythm of the algorithm’s operation.

Yet now, we have perhaps reached a new stage in our technical evolution. No matter how the future unfolds, the emergence of Large Language Models (LLMs) carries a society-shaping force. The expanding use of LLMs raises serious dilemmas regarding their social impact.

Previously, if we bought a computer program, we knew exactly what it was for: it calculated payroll or managed our inventory. It could never happen that the payroll software would suddenly say, “Your employees are underpaid, and this is harming productivity.” But language models are not like that. We do not know their exact location or their scale, and we neither created nor control them. It is a bit like facing a ghost where we do not yet know whether it will help us as an independent partner or remain merely a self-willed, occasionally brooding assistant.

The Significance of Nomenclature

Until now, I have argued that using humanized names like “robot dog” in the fields of robotics and LLM systems is harmful. I believed this was a hidden weapon of marketing departments, helping to blur the line between human civilization and artificial, albeit human-made, systems.

I conducted an experiment and asked my digital assistant (LLM) what names it would suggest that were not of human origin. The result was sobering because I had to realize that for everyday use, we still need the human terms we have come to accept. A “robot dog” is actually a “quadrupedal autonomous terrain platform,” but using this in everyday conversation is obviously impossible. Instead of AI or LLM, we could use the terms “stochastic pattern generator” or “high-dimensional probabilistic matrix model.”

In this article, I deliberately use the term Large Language Model (LLM) instead of Artificial Intelligence (AI) because I believe it more accurately reflects the essence of the technology. However, when I talk about the LLM I use, I call it a digital assistant, which is perfectly accurate. Nevertheless, the experiment showed that a term more suited to a technical tool could still make sense; after all, if artificial intelligence triggers deep-seated fears, I do not believe we would fear for our civilization when facing a mere pattern generator.

How Can We Use It Professionally?

In many stories, there is a wizard or a fairy who solves our problems or protects us from danger with a single wave of a magic wand. Anyone who thinks they have come into possession of such a magic wand is mistaken. We still have a lot to learn and must use it with strict discipline—after all, one must be careful with a chainsaw too.

According to a recent study, even low exposure—as little as ten minutes of use—causes our cognitive abilities to decline while using an AI or LLM; a kind of brain capacity-saving mode takes effect. What would be the solution? Completely cutting ourselves off is the obvious baseline, but that could diminish our competitiveness on both a corporate and personal level.

Our integrity and control can be preserved if we adhere to a few rules, which naturally apply when we enter into joint work as a small business owner or private individual:

• Never enter into a conversation without your own thoughts. Always write down your problem and your own ideas regarding the solution beforehand.

• Always request real internet references for the answers, and actually review them.

• Cross-question the system: what other possibilities exist to solve the problem?

  AdSense

• Warn the LLM if it oversteps its boundaries. If it performs a task you did not request (e.g., generating an image), let it know not to do so.

• Define its scope of operation and establish exactly what your relationship is to it.

Consider the fairy-tale genie: if we ask him to fly us around the world on his carpet, he might do it in the blink of an eye, and we will miss out on the experience we actually desired. If we ask him to bring out his flying carpet and fly us around the world so we can admire it deeply, we might starve to death because he will fly with us for six months and we did not ask for food and water. When we solve a task, we must formulate it as precisely as possible. LLMs do not read our thoughts; they adapt to the nature of them.

Therefore, the LLM is not a silver bullet; it is unlikely to seize power over us. It can write an email text, but it will not send it on our behalf; currently, we make the final decision.

Economic Impacts

Creating an in-house secretary, being annoyed by inconsistencies, and rejoicing when a shopping list is successfully expanded makes for a fun game at home, but currently, the investigation of the complex impacts exerted on the economy raises more critical questions. However, alongside individual utilization, it is even more vital to look at the social and economic impact LLMs will have on our entire society in the future. Debates are already unfolding on multiple levels regarding the effects generated by LLM systems; I will mention just a few:

• The development of LLM systems requires extraordinary capital strength and human resource centralization. This has a dual impact: a reduction in capital availability for other sectors and a drop in productivity due to more expensive or completely unavailable human IT resources.

  AdSense

• Large corporations sometimes look at the profit growth resulting from the use of LLM systems with excessive expectations, but after the introductory period, LLM investors also expect returns, and subscription services have appeared that set limits on unlimited use. This can slow down the speed of learning processes.

• By utilizing LLMs, companies also save human labor, but this is not always successful. Ford reportedly entrusted quality control to an LLM but had to rebuild its team of human specialists.

• The explosive growth in the computing capacity of LLMs must also be serviced. The resulting increase in the price of physical memory components and the emerging shortages could reduce profits in other sectors.

• Increasing computing capacities is highly energy-intensive, which may affect the price and availability of energy.

  AdSense

• As a social impact, the career shift of available labor can be facilitated because, with proper application, a lower entry level can be achieved in certain professions, and this can aid the creation of new jobs.

The LLM can be called a revolutionary product; its full and actual impact on society, the economy, and our own selves remains a secret of the future. Until then, let us learn and get to know the opportunity; perhaps a truly useful tool has landed in our hands, but the rule holds true here as well: humans decide how to use it. Consequently, whether the LLM becomes a curse or a blessing in the future will be determined by the professional and social ethics of human usage.

By: Viktor Szentkiralyi