Albania's prime minister says his country could soon be the first in the world to have an entire government run by artificial intelligence - a move that is inspiring but also deeply concerning.
Last May, European leaders had a little surprise when they visited a conference in Albania: Each of them saw themselves on screen as a baby – speaking in a childish voice and in the politician’s native language. Some viewers laughed when they saw their digital image on stage. Others – like Erdogan, who also had an impressive mustache in his baby form – grimaced.
You could call this story “funny” or “silly,” but you could also say it another way: Albania is trying to do things differently with artificial intelligence, and it’s not ashamed to admit it. And even if some of it seems ridiculous to us, there’s something we can learn from it. Albania is trying to integrate artificial intelligence into all areas of government right now, and its prime minister has even voiced his desire for Albania –
“You will be the first with an entire government made up of AI ministers and a prime minister who will be AI. … That way, there will be no nepotism or conflicts of interest.”
Albania’s love story with artificial intelligence began in recent years, when the prime minister decided it was time to join the European Union. One of the EU’s requirements is that any new member state accept existing rules, laws, and procedures. This is a reasonable requirement, until you realize that it involves a quarter of a million pages in a foreign language. Even advanced countries like Croatia would have had to invest nearly a decade of hard work to complete the process of translating, analyzing, and identifying points of friction between EU laws and national laws.
And Albania? It chose to work with artificial intelligence. From the moment Chat-GPT was launched, the Albanian prime minister understood the potential of the technology. He contacted Mira Moratti – OpenAI’s chief technology officer at the time, and an Albanian by origin – and asked for her help. Moratti immediately agreed.
The result?
"Negotiations with the European Union are being conducted with the assistance of artificial intelligence," the Prime Minister said recently.
He predicts that Albania will successfully complete the process in just five years thanks to artificial intelligence.
This is just one of many uses of AI that Albania is implementing. The country recently launched a website called e-Albania, where it provides 95 percent of all its services online. Citizens who connect to the platform get to talk to… artificial intelligence, of course. The artificial representative helps citizens fill out their tax forms, download various types of documents, pay for services, and more. The system has reportedly saved Albanian citizens – both residents and Albanians working abroad – more than $600 million in the past five years.
Filed your tax returns? Great. Albania is also using AI to review them in real time and identify problems and inaccuracies. It's also relying on AI to examine images from satellites and smart drones, and to detect construction violations and cannabis farms. It even plans to use facial recognition to target delinquent drivers on the roads, and send them an order to slow down while driving. And if they don't slow down? Then the system will also write them an automatic report and send it to them via text message or email.
Given all these uses of artificial intelligence, it is not surprising that the Prime Minister recently stated, according to an article in Politico,
"That technology may soon become the most effective minister in the Albanian government."
It is unclear whether the prime minister has a clear plan for creating that “artificial minister.” He has suggested in broad terms that Albania’s local techies will work on creating an AI model that will be elected as a minister. From there, he predicts, Albania could be the first country whose entire government—from ministers to the prime minister himself—will be made up of AIs. But vision is one thing, action is another, and at least for now, it is difficult to find Albanian developers working in that direction.
But Albania is not the only country in the world, and others are also seriously considering the possibility of replacing human public representatives with artificial intelligence.
But how exactly?
Streamlining or replacement
It is not surprising to discover that artificial intelligence is already helping members of the Knesset and the government do their jobs better. Or at least more efficiently. Just as GPT-Chat saves managers a lot of time in going over documents, translating, analyzing, and making decisions, it could do the same for politicians. This is not a prediction, but a description of the existing situation… at least in some countries.
In England and Denmark, speeches in parliament and the House of Representatives have already been written by GPT-chat. In Brazil, a member of parliament used GPT-chat to draft a new law for the benefit of citizens. The AI wrote the bill for him, and he submitted it “without making a single change, or revealing its source.” The law gained immense popularity and passed without any opposition. Only then did the submitter reveal that it had been written entirely by AI.
In Romania, a new type of artificial intelligence was recently introduced that should make it easier for elected officials to understand and act on citizens’ positions. In Japan, the mayor of Yokozuka launched a large-scale initiative to use GPT chat to produce documents, summarize meetings, and reduce time spent on routine administrative tasks.
And, in Israel too, there is progress in using artificial intelligence to assist in actual work. According to foreign sources (a 2024 report by the US Library of Congress), pilots have been launched in the Israeli Knesset for the use of artificial intelligence tools "to produce summaries of official documents and hearings, to research, analyze, and adapt documents, legal precedents, and academic and specialized literature to the present... and to use artificial intelligence applications to verify the implementation of laws by the government and other relevant public institutions."
Oh, and President Herzog also used artificial intelligence to write a speech for a cyber conference, and won applause.
All of this is almost self-evident. It is “incremental” progress – that is, gradual. Yes, artificial intelligence can help elected officials in their work. But can it actually replace them, as the Albanian prime minister envisions?
Well, not yet. But that doesn't mean there aren't attempts in the field.
In Britain, Steve Endacott created an avatar of himself a year ago and marketed it across the country as a real candidate for the House of Commons. While Endacott made it clear that he would go to Parliament himself to participate in debates and votes, he would act solely on the avatar’s decisions. And how would the avatar make those decisions? Simple: it would constantly consult with voters, answer their questions, and understand what was important to them.
In the United States, Victor Miller tried last year to run an AI for mayor of Cheyenne, Wyoming. Like Endacott, Miller announced that the AI would make all the decisions, and that his role would be to –
"Only to serve as the required human interface… because current laws do not allow artificial intelligence to run independently."
The two political entrepreneurs were disappointed to discover that the public was not yet interested in artificial public representatives. Neither of them was able to get their artificial candidate into politics, although it is worth noting that Miller won almost three percent of the vote. If he had run the candidate in Israel, it is not impossible that he would have succeeded in getting him into the House of Representatives.
These initiatives came from small people, and without receiving support from the political or public system. What will happen when they come from entrepreneurs and developers, who receive support and encouragement from the Albanian Prime Minister himself? Will we see in Albania for the first time an artificial public representative, who also runs for elections on his own?
Maybe, but it's not certain that it will be good for the electorate.
The dangers of artificial selection
It is important for me to emphasize that artificial intelligence, successfully implemented in the political system, can boost the transparency and credibility of every ministry.
But the emphasis is on "successful."
There are many ways in which AI can be used in government in a bad, wrong, or even harmful way. New York City, for example, released a bot that misled residents about the law. In the UK, an AI was launched that was labeled “racist and biased” after it discriminated against visitors of a certain ethnicity. These are well-known cases, and they will likely be repeated many more times in different variations, in any government agency that tries to implement AI.
Another problem is that AI might simply perpetuate the status quo in corrupt government offices. If there is a culture of lying in those offices, it might indulge in that culture itself and amplify it. If the truth is hidden from the public there, it might try to do it itself. Such possibilities seem increasingly plausible as AI advances and we see that it can interpret the instructions it receives in creative ways – for example, to cheat in a game of chess to win.
The third concern is that when AI competes with human politicians – and it already does – it will easily beat them. It will, after all, be able to communicate with citizens and potential voters, and try to persuade them to align themselves with it. We already know that today, the great language engines can be extremely persuasive. We see studies that clearly indicate that AI is on par with the top 1% of humanity in its persuasive abilities. And that is still the case today. With each passing month, AI will become smarter and more advanced. How far away is the day when it can beat any ordinary human politician in self-marketing?
Of course, human politicians will also use AI to promote their public relations. But if it is the 'old generation' politicians, then there is a good chance that AIs – or more precisely, the young, fresh and subversive teams behind them – will succeed in overcoming them. And so what? Who will ensure that the AI we will crown ourselves with is truly accurate and subject to the values of humanity, or at least the values of the voters? All the more so when it can explain itself and all its actions, in the most convincing ways.
These are all important questions for the future, and perhaps even the near future, when the first "artificial intelligence party" is established in Israel. I don't yet know who will establish it and what the platform of that artificial intelligence will be, but I wouldn't be surprised to discover in five years its human representatives in the Knesset.
Success for all of us.
More of the topic in Hayadan:
