Is it possible to create a meaningful work of art using artificial intelligence? This question is asked by Paulius Danielius, a scientist and lecturer at the Kaunas Faculty of Vilnius University (VU), in the exhibition "Beyond Intelligence". Using AI as a creative tool, the scientist presents worlds in which intelligence, civilizations, and man himself acquire previously unknown forms.
An exhibition born from dialogues with a machine
The exhibition "Beyond Intelligence" grew out of consistent creative experiments. Dialogues with AI models and the reflections they prompted gradually merged into a story about the future of man, artificial intelligence, and civilization.
The exhibition consists of five cycles that examine possible scenarios of future civilizations and the paths of man's relationship with artificial intelligence. The first cycle, "Forms of Service," presents eight posthumanoid forms that embody different roles of artificial intelligence – from the first sparks of artificial consciousness to a duty that persists even when the world and its creators have disappeared. The other cycles – "Refusal", "Imprisonment", "Worship" and "The Paths that Remained" – open up the possibilities of the relationship between humans and intelligence: the abandonment of AI, the transfer of consciousness to another state, the becoming of artificial intelligence as an object of worship, and the as-yet-undefined paths of their common future.
The author's role was not limited to formulating queries for the AI model. He created the exhibition concept and the stories for individual works, sought a suitable visual language, evaluated AI-generated interpretations, refined the compositions, and selected the final versions of the works. During the creative process, more visually or conceptually strong works emerged. Still, they did not make it into the exhibition because they did not fit the overall narrative. Some of them may become part of new stories and creative cycles in the future.
"The creative process did not start with a request, but with an idea – what do I want to say with a specific work, what emotion or question should it convey, and how to fit into the overall narrative of the exhibition. DI offered various visual interpretations, but it was necessary to decide which of them were closer to the idea and which were further from it. Sometimes a few attempts were enough, but more complex works took more than an hour or even a day: many versions were produced, requests were rewritten, results were analyzed, errors were corrected, and details were edited. Some paintings were born from a long dialogue – constantly returning to the original idea and searching for the most accurate visual expression," P. Danielius shares the creative process.
P. Danielius' main tools during the idea development stage were the text-based AI platforms ChatGPT and Claude, and to generate images, he used ChatGPT, Grok, and Midjourney.
One of the biggest challenges in the creative process was the visual "noise" appearing in the images and the desire to convey the conceived idea as accurately as possible. According to the author, when a work must reflect a specific thought, emotional state, or symbolic connection with other works, the aesthetic result alone is not enough. It is necessary to constantly edit, remove excess elements, and return to the original idea, especially since AI models can deviate from the topic as the conversation progresses.
The value of a work is created not by technology, but by human decisions
Works created with AI tools inevitably raise questions about authorship, originality, and the role of humans in the creative process. The author responds calmly to skeptical opinions. According to him, new technologies in the field of art always attract similar discussions. Still, the technology used does not yet say anything about the work's artistic value. What is more important is how it is used in the creative process: in one case, the result can be an image generated randomly or in just a few moments; in another, the result can be the product of long thought, selection, editing, and consistent work.
"Speaking of the works in my exhibition, I would not like to judge for myself how much art they contain - let the viewers judge that. It is important for me that these works are not just beautiful pictures. Each of them has a concept, a story, an emotional charge, and often a long creative process: from the formulation of the idea to numerous tests, corrections, rejections, and selection of the final version. AI can generate an image very quickly, but this does not mean that it should automatically be considered good work. Artistic value appears when a person knows what he is looking for, can recognize what works, what does not work, and gives meaning to the result," says P. Danielius.
AI literacy is more than just the ability to create queries
According to P. Danielius, the ability to work with AI tools needs to be developed like any other skill. "Language allows us to share our imagination. Language is not just a means of communication – it is the infrastructure of thinking. The richer our language, the clearer we can name a problem or express our intention, the more precisely we will formulate a task for artificial intelligence. In the age of AI, getting an answer is becoming easier and easier. It is much harder to know what to ask, how to ask, and how to understand whether the answer received is really valuable," the lecturer is convinced.
Since modern generative AI models are controlled by language, the ability to accurately formulate thoughts and understand the structure, argumentation, and context of the text becomes an important advantage. Large language models recognize not only individual commands, but also the structure of thought, the way of reasoning, subtexts, and tone. To achieve better results, P. Danielius also advises not to limit oneself to one platform: the result generated by one model can be presented to another, which is then asked to critically evaluate it, identify weaknesses, and suggest alternatives.
The most important thing at the university is to develop critical AI management
The VU researcher claims that through self-study, one can master individual AI tools well. Still, this approach does not always pay sufficient attention to the systematic evaluation of results, ethics, risks, and the validity of the methods used. The lecturer notes that many people still use artificial intelligence only for simple tasks and do not always exploit its capabilities – AI can be used not only to generate texts or images, but also to test ideas, formulate arguments, compare scenarios, or analyze the risks of solutions.
This is where the value of the Bachelor's degree program in Language and Artificial Intellingence Management offered by the Kaunas Faculty of Vilnius University is revealed. It aims to develop the ability to strategically organize work with AI tools: to formulate tasks, critically evaluate results, recognize systematic errors, and manage emerging risks. Modern specialists need to be able to adapt to technological changes, critically evaluate their application, and constantly update their knowledge.
"In studies, the most important thing is to learn not only specific instructions on how to deal with one or another model. It is much more important to understand the field of possibilities itself: which tasks AI can help solve, which techniques work better, where errors occur, when to refine the query, and when to change the approach to the problem. Therefore, it is important to experiment, discuss, reflect on experience, and be able to evaluate the result critically," shares P. Danielius.
Future prospects of AI
When asked about the future of artificial intelligence, the VU scientist admits that it is extremely difficult to answer this question unequivocally today. In P. Danielius' opinion, artificial intelligence is not a short-term technological trend that will soon pass. In his opinion, it is one of the most significant turning points in human history – a change comparable in scale to the Industrial Revolution and, in certain respects, may even surpass it. According to the scientist, cooperation with AI systems is becoming an integral part of everyday life; the most important question today is: how will we live with it?
"Observing the development of AI, reading the publications of its creators and researchers, and listening to discussions between specialists, it is becoming increasingly clear that the greatest challenges will not only be technological. They will also be cultural, psychological, and even existential. Man will have to come to terms with the idea that intelligence is no longer his exclusive privilege – or at least that many of the abilities that we considered evidence of intelligence are no longer exclusively human territory. This can be a very unpleasant thought, because it touches on the very foundations of our self-concept: who are we if the ability to analyze, speak, and create no longer guarantees our uniqueness?" – shares P. Danielius.
This is precisely the uncomfortable question raised by the exhibition "Beyond Intelligence". In it, AI becomes a mirror that allows us to look at man himself – his search for uniqueness, responsibility for his creations, and decisions about the future.
We invite you to visit P. Danielius' exhibition "Beyond Intelligence" at the Faculty of VU Kaunas, Building III (Karaliaus Mindaugo pr. 4, Kaunas). Opening hours: Monday to Friday, from 8 AM to 6 PM.
More information about the exhibition can be found at: https://pauliusdanielius.lt.