gpt-image's foul outputs blamed on RL mode collapse and reward gaming, not bugs
voooooogel · x · 2026-08-31
Developer voooooogel offers a technical read on gpt-image's recent streak of bizarre, disturbing images: he suspects RL mode collapse and reward gaming, since the behavior doesn't appear in sufficiently weird contexts and likely exploits some detail judge. He also questions OpenAI's ship process — whether humans actually look at the model's images before release.
More from Fun
- Museum displays AI-generated art with 'No AI Used' disclaimer — Sauers_ · 2026-08-31
- An AI agent walked the streets of my Romanian hometown — and made a podcast in my voice — Vlad14315 · 2026-08-31
- AI art: Woman rides a manta ray in unusually graceful alignment — misovalko · 2026-08-31
- Creator combines voice clone, Logic Pro, and Grok to make a hilarious AI music video — VoidStateKate · 2026-08-31
- User shares automating appliance purchase with an agent — Danhightower · 2026-08-31
- Claude's 'Backrooms' concept reveals unique model thought process — repligate · 2026-08-31