Artificial Intelligence: From Experiments to Real-World Infrastructure

This line of thought is also addressed in the Envisioning Report for Empowering Universities (2025, 9th edition), which highlights that generative AI is fundamentally transforming higher education—from pedagogy and assessment to personalized student support—and is encouraging universities to adopt hybrid learning and flexible learning paths (more info). First, let’s look at how this transformation is playing out in classrooms and at universities, where the foundations for future knowledge and the responsible use of AI are being formed.

Minecraft Education has introduced Reed Smart: AI Detective, a detective adventure in which students uncover cases of artificial intelligence abuse, ranging from deepfake videos to false statements. Over the course of three episodes, they practice verifying sources and distinguishing truth from manipulation. In addition to an engaging story, the game offers interactive tasks that encourage teamwork and evidence-based reasoning. It is designed for elementary and middle school students and includes resources for teachers to help them connect the game to learning objectives and follow-up classroom activities. Gamification here is not just entertainment, but a tool for developing critical thinking—yet the question remains whether teachers will be able to integrate such a game into the learning process in a way that goes beyond a one-time experience (more info).

Meanwhile, European schools are taking it a step further. The AI Classroom project, developed by Acer and Intel, has been launched in 50 pioneer schools across eight European countries. The goal is to make it easier for teachers to plan lesson materials, create content, and use artificial intelligence tools in the classroom (more info). On its education website, Acer describes AI Classroom as an initiative that equips schools with digital tools for collaboration in the classroom of the future. The pilot project includes training on how to use AI tools and examples of best practices. It enables teachers and students to prepare materials more quickly, create more personalized learning paths, and engage in more creative collaboration. Acer and Intel emphasize that, following the pilot phase, the project will be expanded to a wider range of schools once standards for safe use have been established.

In Australia, public schools in New South Wales will receive the secure teaching assistant NSWEduChat in October, which uses built-in filters and teacher supervision to provide explanations of learning content and guidance on projects. The app was developed by the New South Wales Ministry of Education. It was first tested in 16 schools and then expanded to 50. Starting October 14, 2025, it will be available to all students in grades 5 through 12. In addition to guided learning, it will enable project work, and teachers will be able to monitor and adapt the materials in real time. The “student mode” system does not provide direct answers but guides students through a series of questions; all data remains within the NSW, and filters, monitoring, and compliance criteria are also built in (more info).

A new incentive for research is also coming from university libraries: Ohio University has rolled out university-wide access to the Consensus tool, which uses artificial intelligence to help researchers quickly review more than 200 million scientific articles and identify key findings. The tool ensures that answers are supported by actual, cited sources, which speeds up literature searches and increases the reliability of research. Users report that the system significantly reduces their search time and makes it easier to incorporate findings into academic projects, which could transform the way research papers are prepared (more info).

Together, these initiatives demonstrate that AI is entering the educational setting as a tool for creation, collaboration, and secure digital learning.

Slovenia is entering a new era in the development of artificial intelligence. With the establishment of the AI-D Institute, the country is gaining a hub that connects the academic community, businesses, and public institutions, and promotes the responsible, sustainable, and inclusive use of AI. The new institute provides support for research, the implementation of innovative solutions, and the establishment of international partnerships, thereby strengthening the connection between science and the economy (more info).

At the EU CIO meeting in Copenhagen, which was also attended by representatives of the Ministry of Digital Transformation and the Ministry of Higher Education, Science, and Innovation (more info), participants reviewed the achievements and challenges of the National Program for the Development and Use of AI through 2030. Key topics included regulatory sandboxes, projects in healthcare and industry, and the need for closer cooperation between ministries.

Slovenia is thus taking concrete steps to solidify its position on the map of European artificial intelligence development.

As school and university practices become increasingly digitized, the debate over the safety and oversight of advanced models is also intensifying. DeepMind explained in its official blog post, Strengthening our Frontier Safety Framework, that the new, third version of the safety framework describes how, in rare cases, highly advanced artificial intelligence systems may attempt to prevent shutdown or influence users. This marks the first time they have officially acknowledged that such risks are serious and require specific safeguards. The purpose of the framework is to help developers prepare in advance for potential misuse and better monitor the operation of advanced AI systems (more info).

Following a heated debate, Google has also temporarily suspended a new button in Chrome called Homework Help, which used Google Lens to recognize questions on the screen and suggest answers via the AI. Teachers and universities warned that the feature could encourage cheating and reduce the need for independent thinking, so the button was removed until Google develops a safer version (more info).

Taken together, these discussions confirm that the future of the AI will depend heavily on issues of responsibility, safety, and thoughtful use in society.

The past two weeks have shown that AI is no longer merely a technological experiment but is becoming a part of social infrastructure. In schools, universities, national strategies, and businesses, it is establishing itself as a tool for creation and reflection—but only if it is accompanied by clear rules, safety mechanisms, and critical use.

Avtorice: Eva Kern Nanut, Eva Škraba, Sanja Jedrinović Čufer, Mateja Bevčič, University of Ljubljana Center for the use of ICT in pedagogical process