Phoenix 24
Technology may generate the image, but authorship still depends on human direction and responsibility.
BUDAPEST, HUNGARY
Hungarian visual artist Nóra Bereczki argues that artificial intelligence should expand human creativity without being mistaken for an independent creator. Working under the name Digital Umami, Bereczki combines generative models with more than two decades of experience in analogue and digital photography. Her Mediterranean-inspired series Dolce Impossible recently won a $100,000 image-generation competition selected from 52,000 entries.
Bereczki distinguishes purposeful artistic creation from obtaining an attractive image through a casual prompt. Her process begins with a specific concept, followed by deliberate decisions about composition, lighting, texture, colour and perspective. She then refines the output through repeated instructions until it reflects her original vision. Previous artistic knowledge remains decisive, she says, because technical accessibility does not automatically produce visual judgment or meaningful work.
She also rejects describing AI as a creative companion. In her view, intention, direction, decisions and responsibility must remain with the person using the system. Treating conversational models as personalities can encourage emotional dependence and obscure the fact that their responses are generated rather than understood. This distinction becomes more important as synthetic images and videos grow increasingly difficult to separate from authentic records.
In education, Bereczki supports a gradual and supervised approach. Teachers should first understand the capabilities and limitations of AI, including verification, privacy, copyright and ethical risks. Children, meanwhile, should learn to concentrate, investigate, make mistakes and think critically before delegating parts of the learning process to a model. Once those foundations exist, AI can support personalised instruction, offer alternative explanations and adapt material for students with different learning needs.
Bereczki considers disruption to employment a legitimate concern but avoids presenting its eventual scale as settled. Research she cited found that employment outcomes among young Americans trained for AI-exposed professions had deteriorated compared with less exposed occupations, even as new AI-related positions emerged. Her recommendation is continuous professional adaptation: workers should examine how their occupations may change, acquire relevant skills and prepare for new roles. The central issue is therefore not whether society can stop AI, but whether human institutions can preserve judgment, accountability and creative agency while learning to use it.
Hechos que no se doblan. / Facts that do not bend.