This series can be traced back to my undergraduate graduation project. Between 2017 and 2022, I recorded a large number of dreams. Over time, certain patterns began to surface—nonlinear connections started to form between them. Back then, I tried to build a "map of dreams" using illustration, and the response was encouraging. But as I kept using the same mode of expression, it started to feel rigid. I had exhausted the illustrative language I knew, and I needed to find a new language—a new crack—to continue unfolding those dreams.

​

My initial turn to AI was driven by something very practical: time. I needed a method that could quickly generate visual material. So I began combining sketches, drawings, and personal image archives, and trained my first set of AI outputs. The results startled me: AI was somehow able to mimic irrational traces—those blurry, fractured, wandering visual rhythms. It wasn’t me controlling it. It was misreading me—and that misreading struck a nerve.

 

I began to realize that AI wasn’t just a tool—it was a blurry mirror. It brought out parts of my mind I hadn’t yet found words for. I don’t see that as a shortcut; on the contrary, it brought me closer to what I was really trying to express.