Summary 🔮
FCK GBRL AI is a creative detour where biomedical image segmentation met the unpredictability of NFT art. The idea was to celebrate the beautiful glitches that emerge from deep learning, not hide them.
The concept 🎭
- Trained a segmentation network on glomeruli scans and curated the outputs that looked visually striking.
- Framed the experiment as an art collection: the artifacts became tokens on OpenSea to challenge the idea of ownership in AI-generated imagery.
- The tone is intentionally irreverent—as the title hints—because the goal was to poke fun at hype while embracing emergent aesthetics.
Reflections ✍️
- Deep models produce unexpected beauty when you let them wander instead of forcing them to behave.
- Collaborating with Gabriel Volpi on the model tuning reaffirmed how critical human feedback is in creative AI projects.