North Macedonia’s AI Politics Problem Is Not the Technology. It Is the Judgment.
By Suad Seferi ·
Artificial intelligence is becoming increasingly visible in North Macedonian politics, but not always in ways that inspire confidence. It is appearing on municipal billboards, in ministry announcements, in political photographs and in social media posts written in the same polished style, usually accompanied by emojis, exaggerated enthusiasm and familiar promises about progress, dedication and a better future. The problem is not that politicians are using AI. They should use it. Public institutions should explore tools that can improve administration, make information more accessible and help civil servants work more efficiently. The problem begins when AI is used as a shortcut around expertise, authenticity and basic human judgment. Recent examples have made that gap difficult to ignore. The new “Welcome to Skopje” billboard, promoted by Mayor Orce Gjorgjievski, quickly became a subject of ridicule after people noticed what appeared to be a five-legged Bucephalus, unfamiliar architecture and a confused synthetic representation of the capital. The mistakes were funny, but the wider message was not. Skopje was being officially represented through an image that appeared to have been generated without proper visual checking, cultural understanding or professional oversight. Artificial intelligence does not understand Skopje. It does not know the meaning of its monuments, the history behind its architecture or the relationship citizens have with the city. It generates images based on patterns, producing something that may look vaguely historical, vaguely Balkan and vaguely monumental without actually representing the place it claims to show. That may be acceptable during brainstorming. It is not enough for an official city billboard. Skopje has photographers, designers, illustrators, architects, historians and cultural workers who understand the city. Choosing a generated image was not the only option. It was simply the easiest one. The Ministry of Internal Affairs recently provided an even more serious example when it published information about the rescue of a foreign citizen near Matka alongside an image that appeared to have been generated by AI. The picture reportedly contained visible abnormalities and did not fully match the circumstances described in the ministry’s own statement. It was later removed after attracting public attention. This was not merely a poor design decision. The Ministry of Internal Affairs communicates about crime, emergencies, police operations and public safety. Its official channels carry authority, and when an image is published next to a real event, citizens reasonably assume that it documents that event or is at least directly connected to it. A synthetic image creates something that looks like evidence without being evidence. The rescue may have happened exactly as described, but the artificial image introduced unnecessary doubt. Was this the real location? Were these the officers involved? Did the operation look like this? Why was an invented scene needed at all? The ministry could have used an official graphic, a map, a clearly labelled illustration or no image. Instead, it chose a visual that appeared to reconstruct reality, then removed it after the public noticed the problems. That should concern anyone who cares about institutional credibility. The same artificial approach is now visible in political language. Posts from ministers, mayors and party officials often follow an almost identical structure: an enthusiastic opening, several emojis, vague praise for teamwork and commitment, followed by a final promise that the institution will continue moving forward toward a better future. Every meeting becomes historic. Every routine activity becomes transformative. Every visit becomes proof of vision, responsibility and progress. The language sounds positive, but it often says very little. Citizens are rarely told what was actually decided, how much something will cost, when it will be completed, who is responsible or what will happen if the promise is not fulfilled. Instead, they receive ceremonial language that has been polished to sound impressive while avoiding anything concrete. Artificial intelligence did not invent political clichés. It simply made them much easier to produce. There is nothing wrong with using AI to improve grammar, organise information, shorten a statement or translate it more clearly. In a multilingual country such as North Macedonia, these tools could help institutions communicate better in Macedonian, Albanian and other languages. But when politicians outsource their voice entirely to a machine, public communication becomes polished and empty at the same time. Citizens no longer hear what an elected official genuinely thinks. They receive language designed to sound professional, safe and optimistic, but without personality, risk or accountability. This is also a poor use of a technology that could provide far greater public value. Politicians and institutions could use AI to analyse legislation, summarise administrative documents, identify contradictions in policy proposals, organise citizen complaints, improve accessibility, prepare multilingual information and reduce repetitive work. It could help ministries understand where public services are failing and assist civil servants in handling large amounts of information more effectively. Those are meaningful applications. Generating an imaginary police rescue is not. Producing a synthetic version of Skopje is not. Making every politician sound like the same enthusiastic chatbot is not. The difference is simple: responsible AI improves the work behind public communication, while careless AI replaces public communication with an artificial version of it. Politicians should treat AI as an assistant, not as a replacement for expertise, reality or their own voice. They can use it to research, organise, compare, translate and prepare first drafts, but the final responsibility must al…