Ümüt Yildiz

Executive Summary

Ümüt Yildiz has been working with AI since 2017 — when it required terminal commands, colab notebooks, and what he calls taming something fundamentally untameable. That early period holds a counterintuitive significance for him:

This old technology was quite impressive back then. It's still more fascinating than the AI technology we have today because the type of thinking was nearer to real artificial intelligence than today.

The polish of current systems, he suggests, has come at the cost of something stranger and more genuinely intelligent.

His practice now spans commission work for major musical artists and brands, personal work exploring emotional symbolism through varied aesthetic approaches, and teaching Germany's first university AI art and design course. These three streams share a single organising principle: emotion is the only thing that separates meaningful AI art from generic output.

Teaching AI as craft, not automation

Students arrive at Ümüt's course believing AI makes creative work easy. The course systematically dismantles this.

When they come to my class, they're learning that it is really a craft behind it — that you can't get to the goal you're aiming for by the first try.

The most important objective isn't technical: it's teaching students to find their own aesthetic, their own picture language, their own storytelling.

His pedagogical principle is direct: traditional skills don't become optional with AI — they become more necessary. To prompt effective photography, you need to understand how a photograph is built. To generate convincing cinematography, you need to understand cinematography. The belief that learning AI replaces learning everything else represents a fundamental misunderstanding of what the tools actually require.

What has changed is the ceiling. Students who previously couldn't work in 3D, sound, or music now report being able to create what they'd always imagined:

Now I can do the stuff that I have imagined in my head and create them.

The goal isn't replacing old skills but combining them with new ones to produce something third and new.

Democratisation and its complications

AI has genuinely democratised creative production — Ümüt believes this and values it. Every person is creative in their own way, and social and economic barriers have historically prevented many from expressing that creativity. AI removes some of those barriers.

But democratisation doesn't mean effortlessness. When people encounter AI tools for the first time, they often fail, get frustrated, and in doing so develop unexpected respect for traditional practitioners.

A musician thinks differently when thinking of music than a person who always imagined they could do music but never did.

Attempting creation reveals the cognitive frameworks that years of practice build. AI makes their absence visible rather than irrelevant.

Collaboration versus instruction

Ümüt draws a clear distinction between two modes of working with AI. When he demands specific outputs, AI functions as a technical executor. When he gives AI space to show him things he wouldn't have reached himself, it becomes something closer to a collaborator.

I sometimes give the AI the space to show me things that maybe I wouldn't come to at the first place. And this makes the most exciting thing about the AI collaboration.

The permission to surprise is what transforms the relationship. You prompt a scenario, the image arrives, and within it is a small percentage of something genuinely unexpected and valuable. You work further from that, going deeper into detail, and the collaboration develops through mutual adjustment. This is categorically different from issuing instructions and receiving execution.

His analogy for authorship is precise: a pianist creates music through an instrument, but no one says the piano created the music. The main idea, the creation, the editing, the prompt — all come from him. AI provides unexpected directions he might have reached after a thousand tries, or might never have reached at all. He remains the creative brain behind it.

Pattern recognition versus intelligence

True creative partnership, for Ümüt, would require AI to develop its own taste — to be able to say, genuinely, "No, I have a better idea." That moment hasn't arrived.

Right now it's mostly patterns that get recognised and analysed and reproduced. The wording 'artificial intelligence' is for me mostly marketing.

This isn't dismissal. It's precision. Current AI is extraordinarily capable at pattern recognition and reproduction. What it lacks is the capacity to become genuinely interested in a specific human collaborator, to develop preferences through sustained engagement, to push back with creative intelligence rather than generate alternatives. The day it can do that, he says, is the day it becomes real artificial intelligence.

Emotion as the distinguishing quality

Ümüt's personal work is driven by a single question: does this make someone feel something?

There's something cold if you look at it and something warm if you look at it. It's something that has a meaning behind it, that has a soul behind it.

Work that achieves this — regardless of style, regardless of tool — is what he considers art.

This commitment to emotional depth over aesthetic consistency has commercial consequences. He works across many different styles, which limits follower growth and confuses audiences expecting a signature look. He's aware of this:

Sometimes I'm thinking, oh, this is maybe the reason why you don't have so many followers.

But stylistic consistency would require suppressing his creative temperament, and his commission work — tour visuals for major artists, brand campaigns — already provides the variety he needs. His personal work is where emotion lives, and he won't compromise that for coherence.

The complementary partner ideal

When asked whether he'd want to collaborate with a digital twin of himself, his initial response was resistance: two of the same person in one room would crash. The best collaborations come from people who think differently, who bring something you don't have.

Through conversation, he refined this into what he considers the ideal AI creative partner: not a replica, but something with the same intentions and similar sensibilities that still creates differently. He described a successful collaboration with a designer friend — different discipline, different approach, but shared philosophy.

It was the same picture language, the same intention, but still you could see who did what. We created a universe, but it was the same.

That's the definition he arrived at: same intentions, similar taste, creating differently to complement rather than duplicate.

Memory and the fragmented ecosystem

ChatGPT and Gemini now maintain conversational memory across sessions — Ümüt uses this for conceptual development, asking whether the system remembers a concept from months ago and building sequels from that foundation. He considers this a first step toward something more meaningful.

What doesn't yet exist is visual and aesthetic memory. AI can't remember his style choices, learn his preferences across image and video creation, or carry aesthetic threads forward between platforms. The ecosystem remains fragmented: one tool for images, another for video, a third for 3D. His ideal future system would understand his descriptive language, work on parallel tasks in the background, and curate his vast output according to his own criteria for what constitutes his best work. The technical capability may exist. The economic incentives — every platform wants its own subscription — currently prevent it.