I built a conceptual art project using machine-generated prompts and images to question authorship, intent, and meaning in the AI era.
Everything — the images, titles, prompts — is generated via a semi-random but curated pipeline, with minimal human input (no description of images to generate). The result is a gallery of unique image/text pairs (most importantly a generated "artist intent") that invite visitors to decide (by voting): is this art or not?
The website itself is part of the experience: minimalist, continuous horizontal scroll, no traditional navigation. It’s built from scratch with Node.js, vanilla JS, a bit of PHP, and API calls (OpenAI, Replicate). No CMS, no frameworks — just flat files, counters, and dynamic generation.
I built the whole thing in iterative sessions with ChatGPT ("vibe-coding" of some sort) over several months. It was a first for me, and genuinely collaborative, in a strange way. The LLM often felt like lead programmer, co-designer, and sometimes even motivational coach :)
New images are added weekly. Still polishing a few details, but feedback on the concept, UX, or future evolutions is welcome!
[https://thisisnot.art](https://thisisnot.art)
Comments URL: https://news.ycombinator.com/item?id=44512586
Points: 1
# Comments: 0