Aký bol rok 2025 pre AI a čo nás čaká?

As the year comes to an end, it is worth taking a moment to look back. 2025 was an exceptional year for modern technology and artificial intelligence. Things that, until recently, belonged to visions of the future have quietly become part of our everyday lives.

At KInIT, we consider technology not only through research, but also in terms of its impact on people and the world around us. With this in mind, we asked our experts how they viewed 2025. What caught their attention, what concerned them, and where they think we are heading next.

Which trends dominated in 2025 in your opinion?
According to the Cambridge University dictionary, the word of the year 2025 was „parasocial“, and chatbots based on generative AI played a big role in this. I would add to this further societal division, the increase in the spread of fictions, and the dehumanization of people through deepfakes. On the other hand, I also perceive the beginning of awareness of the risks of chaos, which floods us especially through social media and politics. The world is full of contradictions, and the truth becomes tiring.

What can we expect in the new year or the near future?

Advances in artificial intelligence are sure to come, and they can improve our lives in many areas. The question remains, however, how much of our humanity can be appropriated by technology and people with authoritarian tendencies. I don’t dare to guess that today.

Which trends dominated in 2025 in your opinion?
A less visible but very important trend is the creation of native European, public, and transparently developed large language models, such as the German SOOFI. Although they are currently smaller than the largest commercial models, their existence is crucial. They can form the foundation for future solutions that will align with European values and standards.

What can we expect in the new year or the near future?
Predicting the future is not very advisable, but within the European Union, it will be important to follow the implementation of the Digital Services Act. The first requests for access to large platform data are currently being created, and in a few months we will see what this tool can really enable.

Which trends dominated in 2025 in your opinion?
From a regulatory perspective, the discussion focused on how to realistically implement new frameworks such as the AI Act or the Digital Services Act, especially for generative AI and recommendation systems. Strong lobbying by major players led to pressure to soften requirements even at the European Commission level.

What can we expect in the new year or the near future?
The debate on AI regulation in Europe will continue in 2026. We will see further disputes around transparency, fines, and attempts by large platforms to avoid obligations. I expect pressure to weaken regulatory requirements to increase.

Which trends dominated in 2025 in your opinion?
The most significant progress was in agentic AI and video generation. At the same time, there is increasing discussion about Edge AI, meaning processing data directly on devices, which increases security and speed. Discussions about the trustworthiness and explainability of AI are also very important, especially in fields such as medicine.

What can we expect in the new year or the near future?
Further development of large language models is certain, but attention will also shift to working with time series, for example in energy, finance, or biology. An interesting direction is Time Series Foundation Models, which help us better understand the concept of time. Additional regulatory initiatives in AI will also be important.

Which trends dominated in 2025 in your opinion?
Agent systems and their integration into real processes, the standardization of protocols like MCP, AG-UI, or A2A, and the development of multimodal AI. These trends make it easier to connect AI with production systems and use advanced services even without strong research backgrounds.

What can we expect in the new year or the near future?
Continuation in these directions, with a stronger emphasis on benchmarking and evaluating the capabilities of models and agents. An open question remains possible deregulation of AI in Europe. A significant challenge will be distinguishing between content created by humans and by machines, which will be an even more urgent topic in 2026.

The year 2025 has shown us that artificial intelligence is becoming an increasingly integral part of our lives. At KInIT, we therefore strive to conduct research that is close to reality and responds not only to new possibilities but also to the questions and risks that AI brings.

This year has shown us the power of agentic and multimodal AI, the importance of Edge AI, the growing need for regulation and transparency, as well as the challenges related to anthropomorphization and parasocial phenomena. We see that technologies change our everyday life, but at the same time remind us that the purpose of research lies in understanding their impact and connecting it with the world around us.

Therefore, we want to continue asking questions and seeking ways to make technologies understandable and truly useful. We are glad that the topic of artificial intelligence resonates with you and that we are not alone in this.

Thank you for being with us.