Artificial intelligence is changing the way we think, learn and work

In the article How AI is quietly distorting academic enquiry – and what to do about it, the author points out that generative AI is increasingly changing the ways in which students and researchers search for information, formulate research questions and acquire knowledge. He particularly highlights the risk that, rather than reading sources in depth, users are increasingly turning to AI summaries and generated answers, which can lead to more superficial research and less critical evaluation of information. Furthermore, AI often presents primarily the most likely or prevailing answers, which can affect the diversity of research approaches and inadvertently steer researchers towards more predictable or already established interpretations.

Similar challenges are also highlighted in a news article on the use of generative AI in the preparation of literature reviews, which emphasises the risk of fabricated, inaccurate references (hallucinations), incorrect summarisation of sources and a loss of information traceability. The article presents several specific approaches for the safer use of AI in research work, including verifying references, using a variety of sources and tools, and clearly documenting the use of AI and the literature search process.

There are also increasingly frequent warnings that over-reliance on AI can lead to so-called ‘cognitive debt’, whereby the need for mental effort – which is key to the development of understanding, argumentation and critical thinking – is reduced. The article Think again: reclaiming our ‘cognitive debt’ from AI points out that AI is increasingly taking over tasks such as summarising, organising ideas, writing and finding answers, which may cause users to gradually lose the habit of independent thinking.

At Princeton University, due to the widespread use of AI and difficulties in detecting unauthorised assistance, they have amended their 133-year-old honour code system. Until now, students have sat exams without invigilators, as the system was based on trust, personal responsibility and the duty of students to report cheating by their peers; from July 2026, however, exams will be invigilated in person by teaching staff. It is interesting to note that the changes were called for not only by teaching staff but also by students, who pointed out that the use of personal devices and AI makes it harder to detect breaches, whilst reporting suspected cheating is linked to social pressure, anonymous reports and a fear of being singled out.

The study Artificial Intelligence and Grade Inflation also opens up an important new dimension to the debate on assessment. The author analysed more than 500,000 marks at one of the larger research universities in the US over the period 2018–2025 and found that, following the release of ChatGPT, marks had risen noticeably in subjects with assignments more exposed to AI, such as writing and programming. The proportion of marks increased by approximately 30 per cent compared with the situation prior to 2022, with the effects being greater in courses where homework carried greater weight. The author points out that generative AI may reduce the informational value of marks, as part of the assessed work is carried out by the tool (rather than the students).

The article Why AI literacy belongs in the first-year experience advocates for the inclusion of AI literacy as early as the first year of study. The author emphasises that, regardless of their field of study, students should understand the basic principles of how generative AI works, its limitations, risks and impact on the reliability of information. AI literacy is thus increasingly becoming part of general academic literacy, much like information or digital literacy.

The active use of AI is also encouraged by an article describing an initiative at the University of Virginia, where the aim is to develop AI literacy primarily through practical application. Instead of separate theoretical modules, they propose integrating AI tools into students’ day-to-day work, with an emphasis on experimentation, reflection and understanding the limitations of the technology. Students are expected to learn about AI primarily through its use and the critical evaluation of results, not merely through descriptions of the technology.

A similar line of thinking is also evident in the article Dear educators, Gen Z here. Could you please teach us like it’s 2026?, in which a student points out that Generation Z, having grown up in a digital environment, expects different forms of teaching than previous generations. In her view, students do not simply need prohibitions or warnings regarding AI, but rather help in understanding how to use the technology meaningfully, responsibly and in relation to real-world problems. The article emphasises that students are already using AI to organise their work, study and write, whilst universities are not always keeping up with the pace of change.

The fact that students are becoming one of the key user groups for generative AI is also demonstrated by the new initiative OpenAI ChatGPT Futures: Class of 2026. The programme gives students access to advanced AI tools and encourages them to explore their use in learning, creating and developing ideas. At the same time, it raises the question of how universities should integrate such tools into the learning process in a way that supports learning rather than merely productivity.

Higher education institutions continue to move away from isolated experiments with AI towards system-wide policies, guidelines and strategic management. A good example is the State University of New York (SUNY), which comprises 64 campuses and has adopted a system-wide policy on the use of AI. These regulations provide for the wider use of AI in teaching, research and student support, whilst also requiring safeguards for data privacy, checks on the bias of tools, the retention of human judgement in important decisions, and training for students and staff on the safe, ethical and effective use of AI.

The Coursera report on AI in higher education highlights the gap between the rapid adoption of AI and institutional readiness. A survey of 4,200 students and academics across five countries shows that 95 per cent of respondents already use AI in their academic work; however, according to the academics, only 26 per cent of institutions have formal guidelines on the use of AI. The report therefore emphasises that widespread use does not in itself constitute a strategic advantage; universities must establish clear rules, invest in AI literacy and address issues of academic integrity.

The article Half of Campus Tech Leaders Question AI’s ROI describes a survey conducted among university heads of information and technology services. Half of those surveyed believe that the return on investment in AI is still unclear or lower than expected, whilst only 29 per cent believe that the investments have met or exceeded the expected return. In this regard, increased staff productivity is most frequently cited as the greatest tangible benefit of AI, whilst improvements in teaching and learning are mentioned considerably less often. Among the greatest risks for higher education institutions, respondents cite difficulties in recruiting and retaining IT staff, cyber threats, and rising or unsustainable costs.

The European Parliament and the Council of the EU have reached an agreement on simplifying the implementation of the Artificial Intelligence Act (AI Act). The agreement is designed to facilitate the implementation of the rules, promote innovation and reduce administrative burdens, whilst maintaining a focus on safety and the protection of fundamental rights. This is particularly important for universities because the use of AI is increasingly being integrated into a broader regulatory framework, which will influence the development, testing and use of AI systems in educational and research environments.

The University of Central Florida has presented the redesign of online courses using AI, in which lecturers use generative AI primarily to check the alignment of learning objectives, assignments and standards, identifying potential gaps, redesigning learning activities, preparing draft materials, formulating questions for discussion and adapting content to different groups of students. The article emphasises that AI should not replace teaching work, but rather reduce routine tasks and free up more time for working with students and planning learning experiences. It is highlighted that, when using AI, teachers must retain control over course objectives, the quality of content and pedagogical decisions.

The question of how to meaningfully integrate AI into STEM subjects is addressed in the article The Real Problem With AI in STEM Education. The author points out that the problem is not merely that students use AI to solve tasks without critical reflection, but that, due to poorly designed learning activities, they may lose sight of their understanding of fundamental concepts. In his view, educators should place less emphasis on prohibiting the use of AI and more on designing tasks that require the explanation of procedures, the evaluation of results and the application of knowledge in new situations.

The question of how AI will affect work and which skills will be most important in the future is becoming an increasingly significant issue for universities too. A professor at Imperial College Business School points out that no job is completely safe from the impact of AI, although some professions are more resilient than others. Among the key skills of the future, he highlights the ability to collaborate with AI, critical evaluation of results, innovation, interpersonal skills and adaptability. In his view, the most successful individuals will be those who know how to use AI to support their work, rather than simply automating existing tasks.

Similar changes are also highlighted in the article Beyond skills: Why story will define survival in the AI age, in which the author emphasises that, in the age of generative AI, technical skills alone will no longer suffice. The ability to create meaning, interpret and connect information, and understand human experiences and social context is set to become increasingly important. The article highlights that universities will no longer prepare students merely to carry out tasks, but increasingly to understand complex social and organisational situations and to develop reflective judgement.

The question of AI’s impact on working environments is also becoming an important area of research. Stanford HAI has announced the establishment of the ‘AI and Organisations Lab’, which will investigate how artificial intelligence is transforming work, organisations and collaboration. Researchers aim to gain a better understanding of how AI affects productivity, decision-making, organisational culture and inequalities in the workplace.

RTV Slovenija reported on a study which found that an increasing number of young people in Europe are turning to artificial intelligence for emotional support. According to the study’s findings, around half of young people use AI chatbots to discuss personal problems, feelings or stress, whilst some even regard AI as a “friend”. Although AI can provide a sense of support, it cannot replace professional psychological help, human empathy and genuine interpersonal relationships. Experts therefore warn of the potential for emotional dependence on AI, privacy concerns and the risk that, when facing more serious mental health issues, young people might seek support primarily from digital systems rather than from professionals.

OpenAI has also introduced updates to the ChatGPT system to better recognise risky and sensitive situations in conversations with users. The aim of the changes is to improve responses in cases where users mention self-harm, mental distress or other sensitive topics. The announcement highlights the importance of context and caution in the development of systems that people are increasingly using for personal conversations and support.

In Slovenia, issues relating to the responsible development of AI were also at the forefront of the President’s Forum on Ethical Artificial Intelligence, where participants emphasised the importance of developing trustworthy, transparent and human-centred AI systems. Among other things, the discussion highlighted the need for greater digital literacy, responsible technological development and the involvement of various disciplines in shaping future rules and practices for the use of AI.

The fact that AI is increasingly making its way into the physical spaces of cities and services is illustrated by the news of the arrival of Baidu’s robotaxis in Ljubljana. The project represents one of the first examples of autonomous vehicle testing in Slovenia and raises questions regarding cities’ technological readiness, legislation, safety and user trust in autonomous systems.

The environmental impact of AI development is also receiving increasing attention. The programme ‘Green Friday’ highlighted that the largest data centres already consume more energy than many smaller towns, and that further growth in the use of generative AI will only increase these demands. The debate focuses primarily on issues of energy efficiency, water consumption for cooling data centres, and the search for more sustainable approaches to AI development.

OpenAI has introduced the ability to use the Codex tool on mobile devices, which enables programming and the development of software code using natural language or voice commands, even for users without in-depth technical knowledge. The article emphasises, in particular, the simplification of software development and the expansion of access to programming tools, with the UI increasingly taking on the role of an assistant in writing, correcting and interpreting code.

An increasing number of discussions are also raising the question of AI autonomy. The article When Artificial Intelligence Begins to Create Itself examines the development of systems capable of independently improving their own processes or assisting in the creation of new models. In this context, the author highlights both the potential for faster technological development and the issues of control, security and accountability in the development of increasingly autonomous AI systems.

Another significant trend is the development of multimodal artificial intelligence systems, which combine text, images, sound and other types of data. An article on multimodal data emphasises that it is precisely the integration of different types of information that is becoming one of the key directions for the development of the next generation of AI, as it enables a more complex understanding of the environment and broader possibilities for application in research, industry and everyday life.

At the World Economic Forum, it was pointed out that the future of AI may not be based primarily on ever-larger models, but on smaller, specialised and more energy-efficient systems tailored to specific tasks (SLMs – small language models). Such models are expected to be cheaper to develop, easier to use and more accessible even to smaller organisations and countries.

If you have tried out an interesting use of AI in education at your faculty, as part of a course or within the context of your professional work, we warmly invite you to share your example with us via the form for collecting examples of good practice. We look forward to reviewing the submissions and including selected contributions in future issues of the UI newsletter, so that together we can build an overview of useful and well-considered approaches within the university environment.

Authors: Maja Kosmač, Sanja Jedrinović Čufer, Mateja Bevčič, Eva Kern Nanut, Eva Škraba, University of Ljubljana Centre for the Use of ICT in the Teaching Process