Four months ago, Prime Minister Kristen Michal hosted Arthur Mensch, the CEO of Mistral AI, Europe’s most important artificial intelligence company, at Stenbock’s house. After the meeting, they were optimistic, they talked about cooperation plans to make the country more efficient in education and the defense industry, and Estonia’s new artificial intelligence jump. The meeting was said to be fruitful and at least partly aimed at supporting European technological sovereignty.

    At a recent meeting with entrepreneurs, Prime Minister Michal admitted that artificial intelligence has become a daily professional assistant for him, without which it would be difficult to keep up with the pace. Prompted by this, Digigenius asked all members of the government which artificial brain models they use.

    The answers were surprising and reveal the fact that the entire cabinet is firmly leaning towards the US tech giants. None of the Estonian ministers currently use Europe’s own flagship Mistral solutions, leaving Ränioru’s algorithms as the sole ruler in the corridors of state management.

    The State Chancellery is a believer in AI

    Technology for competitive advantage

    Prime Minister Christian Michal did not find time for several weeks to answer our inquiry, and Jürgen Randma, the communication specialist of the State Chancellery, commented on the use of AI on his behalf, and in his answers, in addition to the use of the Prime Minister’s artificial voice, he opened up the background of the State Chancellery’s introduction of AI in general.

    According to Randma, the introduction of artificial branch solutions was a joint decision of the State Chancellery’s management to offer employees the opportunity to test new technology and achieve greater work efficiency. In practice, artificial intelligence is primarily used for structuring one’s thoughts, making summaries of longer texts, checking language and style, finding information and data analysis.

    “Perhaps an artificial intelligence can help structure one’s thoughts and edit the text, but political positions, choices and wording are always the responsibility of the prime minister himself. Although sometimes it may seem that an artificial intelligence “does the work for us”, the reality is rather the opposite,” said Randma. According to him, in order to get a good result, a person must think very precisely, ask, check and take responsibility.

    In terms of security, the State Chancellery has clear principles. “If the material is not intended for public presentation, it is also not suitable for discussion with artificial intelligence solutions. These principles apply to all state institutions and officials. Each state institution decides for itself which artificial intelligence solutions to use in its work, based on the current security and data protection requirements,” added Randma.

    The State Chancellery does not have a separate artificial branch that is completely disconnected from the Internet and works in a so-called sandbox. All employees have access to OpenAI’s ChatGPT paid service, which reportedly does not use agency data to train the overall model. In addition, all employees have the opportunity to use the free version of Microsoft Copilot.

    “ChatGPT was introduced in the State Office 1.5 years ago because at that time it was the best generative artificial intelligence solution provider based on initial comparisons. From April 2025, the possibility to conveniently use Microsoft Copilot from work devices was added,” he concluded.

    Wall to wall usage patterns

    Foreign Minister Margus Tsahkna uses artificial intelligence mainly to extract necessary information from the public space and to get quick overviews of the most important news of the day. His use of artificial eyes is said to have come from a personal desire. “I use ChatGPT on a daily basis. I’ve also tried CoPilot, a Microsoft alternative, but I’ve always come back to ChatGPT,” Tsahkna described.

    Minister of Education and Science Kristina Callas has used artificial intelligence himself both in politics and in his work as a lecturer at the University of Tartu. His ministry has enabled workers to use the paid version of ChatGPT.

    “ChatGPT 5 is the strongest of the free language models on the market today. I have also tried other solutions, especially Google Gemini, but I gave it up the moment I started using the paid version of ChatGPT. Professionally, I use the language model primarily to support thinking,” he said.

    This helps Kallas in preparing policy papers, comparing policies and solutions of different countries, and gathering information quickly. “I also use artificial intelligence to structure ideas when preparing presentations and speeches,” added Kallas.

    However, according to Kallas, he writes all his speeches himself. As the reason, he points out that ChatGPT’s knowledge of the Estonian language and culture is not yet at a level that would allow the use of ready-made conversations. “Therefore, the ordinary person does not see direct “traces” of artificial intelligence in my work or in my speeches,” he noted.

    In addition to his main work, he also uses artificial intelligence in the preparation of his studies and has also encouraged others to experiment and consciously use artificial intelligence.

    “I am still a lecturer at the University of Tartu, and I recently taught a history lesson as a minister, for which I used ChatGPT to prepare worksheets and study materials. Artificial intelligence does not replace a teacher or a decision-maker, but when used correctly, it helps to save time and sharpen thinking,” added Kallas.

    More questions than answers

    Minister of Energy and Environment Andres Sutt believes that artificial intelligence is an irreplaceable help in situations where colleagues have more responsibilities than time and questions constantly arise. He sees chatbots as a valuable alternative to traditional search engines, preferring the paid version of ChatGPT, although he has also used Claude and Gemini to bounce ideas around.

    “In most jobs I’ve experienced in my life, every day offers more questions than answers. Every answer tends to be followed by two more questions. There is always a limited amount of colleagues and acquaintances with whom to exchange ideas, and everyone has more responsibilities than time. Tireless, although chatbots with their own quirks can sometimes partially replace them,” said Sutt.

    Sutt confirmed that there are probably no signs of the use of artificial intelligence in his speeches and texts, if only very superficial signs. According to him, the reason lies in the fact that even if you try to let robots dictate the entire text, it has not been of high enough quality.

    “The most practical way has been to ask robots to critique your texts and perhaps suggest better and shorter wordings. A robot editor. And of course chatbots have been useful for quick information searches. There have been talks about the worsening problems of standard Google search for years, but even without these problems, artificial intelligence would often surpass the old search engines,” said Sutt.

    According to him, you have to take a critical view of information from foreign pages on the Internet anyway. And here he does not make a difference whether this information was found by a person using the Internet or by some artificial intelligence.

    Sutt considered it necessary to point out that in the game of chess, a computer beat a person almost 30 years ago, but a person in cooperation with a computer has been able to beat all kinds of opponents. Based on this, he draws a parallel with the use of artificial intelligence and finds that refusing the help of language models is not a good idea.

    Your brain is trained enough

    This is where the drastic difference between Suti and the Minister of Infrastructure emerges Golden Leisi in use of an artificial branch, the latter of which directly states that he does not use the help of any artificial branch.

    “I’m old enough that my brain has been sufficiently trained and I don’t use artificial intelligence in my work. The advantage of being a leader, including a minister, is that you have a strong team around you. Before any bill or other material reaches my desk, a lot of work has already been done,” said Leis.

    According to Leis, the background, choices and effects have gone through several rounds with the ministry’s experts, lawyers and analysts as well as his own advisers. “From there, my job is to make a choice and take responsibility: does it really solve the problem and is it fair. I have to be responsible for that, not the machine,” added Leis.

    The Ministry of Justice and Digital Affairs has its own artificial branch

    Minister of Justice and Digital Affairs Lisa from Pako uses artificial intelligence as a systemic aid to improve the quality of law-making, while he, as a private individual, pays for the paid version of ChatGPT.

    On his desk, an artificial branch is indispensable for analyzing drafts, helping to manage the huge volume of the European Union’s legal space and find possible contradictions there. Pakosta emphasized that technology is a necessary helper for him.

    “In addition to ChatGPT, I also use an internal artificial intelligence-based draft assessment tool created by the ministry itself. I emphasize that I only use them to process documents that have already become public – I do not enter information intended for internal use in the institution,” said Pakosta.

    He also uses artificial intelligence to edit texts so that the laws are clearer to people. “I view the artificial limb as an additional aid: it is effective, but it can be wrong, so the results are always checked by a person,” concluded Pakosta.

    Minister of Culture Heidy Purga sees artificial intelligence as a useful assistant in analyzing information and making summaries of voluminous documents. For him, the technology is primarily used when it is necessary to compare different materials while maintaining a strict boundary with confidential government materials.

    Minister of Economy and Industry Erkki Keldo sees artificial intelligence as a smart helper in routine tasks, but he has so far used it rather little in his ministerial position. Since the daily work of the Ministry of Economy and Communications (MKM) is done with Microsoft tools, Copilot is also the main choice, with which experts work through information materials and look for references to sources.

    “The introduction of AI is one of the most important trends in development both in the private sector and in public administration, in order to be more efficient and innovative. The ministry’s experts prepare comprehensive analyzes and reviews for me, so the necessary preliminary work and simplification have already been done to reach my desk. Tehisaru, however, is a great help in gathering thoughts on unfamiliar topics and making quick summaries,” said Keldo.

    Minister of Regional Affairs and Agriculture Hendrik Johannes Terras uses artificial intelligence as a working tool very little and does not indicate a specific preference for any platform. According to him, the use of digital solutions in the ministry is not an end in itself, but a need-based aid in places where it is essentially appropriate. “Artificial intelligence is primarily useful in the stages of routine text work, but it does not eliminate the need for human input,” he noted.

    Instead of Social Minister Joller, AI answered the questions

    Unlike some of his colleagues, the Minister of Social Affairs brought Karmen Joller pointed out as an interesting nuance that he also used the help of an artificial intelligence in preparing the answers to this request, but he also emphasized that the end result must always be polished by a person, as he does not tolerate robotic language.

    Joller uses ChatGPT as a daily tool and, according to himself, has already trained his AI quite well, but despite this, he always rereads the results critically, as the machine sometimes creatively manipulates the facts. He noted that his initiative was personal, getting help from an acquaintance at the very beginning.

    “The OpenAI solution is currently the most flexible and understands well the Estonian-language context, including politics and healthcare. I use it both to structure texts, to organize thoughts, and to think through some ideas when time is up,” described Joller.

    “Although I’ve tried other models, ChatGPT has remained the main tool for convenience. It saves me time and forces me to be more precise without replacing experience or responsibility,” he added.

    In the Ministry of Defense, artificial intelligence is a strategic inevitability

    Despite Digigenius’s inquiries, the Minister of Finance still owed an answer regarding his artificial intelligence habits Jürgen LigiMinister of the Interior Igor Taro and the Minister of Defense Hanno Pevkur.

    On behalf of the Ministry of Defence, a general comment was provided by the spokesperson of the ministry, according to whom a common requirement has been set for all employees that the artificial intelligence may only be used for processing public information, and information for internal use (marked as AK) shall not be entered there.

    “Tehisaru’s help can be used, for example, to prepare press releases or statements, while a person always checks this work,” said the spokesperson of the ministry.

    https://i.redd.it/7tc69u5iv1lg1.jpeg

    Posted by railnordica

    Share.

    8 Comments

    1. Tundub et et kõneisikud võiks küll AI-ga asendada.

      “Selgita minu laiskust ja tegevusetust, aga heast küljest”….ENTER….

    2. Distinct_Trash7491 on

      Huvitav millal opositsioon AI-d kasutama hakkab:

      “Palun võta minu mõtted ja kirjuta need ümber päris faktide ja väärika sõnavaraga”

    3. Abhorrent_Appeal_745 on

      Respect Leisile, muus osas on see aga paras õudusunenägu. See on vastuvõetamatu, et osad ministrid lasevad chatbotil enda eest infot otsida ja seda infot analüüsida/võrrelda. See saab ju mingite teadmata parameetrite järgi valida, mis infot see kasutajale edastab, rääkimata selle info analüüsimisest ja tõlgendamisest.

      USA korpode loodud mudelid on tõenäoliselt kallutatud nende korpode väärtushinnangute poole ning nad võivad mistahes hetkel vabalt neid mudeleid veel rohkem kallutada, et teiste riikide poliitikat mõjutada. Kas see oleks okei, kui meie ministrid võtaks endale näiteks Sam Altmani “erapooletuks” nõunikuks?

    4. Las AI võtab kõik juhtivad ja loomingulised rollid üle, et meie saaks keskenduda raskele füüsilisele tööle.